The Science and Information (SAI) Organization
  • Home
  • About Us
  • Journals
  • Conferences
  • Contact Us

Publication Links

  • IJACSA
  • Author Guidelines
  • Publication Policies
  • Digital Archiving Policy
  • Promote your Publication
  • Metadata Harvesting (OAI2)

IJACSA

  • About the Journal
  • Call for Papers
  • Editorial Board
  • Author Guidelines
  • Submit your Paper
  • Current Issue
  • Archives
  • Indexing
  • Fees/ APC
  • Reviewers
  • Apply as a Reviewer

IJARAI

  • About the Journal
  • Archives
  • Indexing & Archiving

Special Issues

  • Home
  • Archives
  • Proposals
  • Guest Editors
  • SUSAI-EE 2025
  • ICONS-BA 2025
  • IoT-BLOCK 2025

Future of Information and Communication Conference (FICC)

  • Home
  • Call for Papers
  • Submit your Paper/Poster
  • Register
  • Venue
  • Contact

Computing Conference

  • Home
  • Call for Papers
  • Submit your Paper/Poster
  • Register
  • Venue
  • Contact

Intelligent Systems Conference (IntelliSys)

  • Home
  • Call for Papers
  • Submit your Paper/Poster
  • Register
  • Venue
  • Contact

Future Technologies Conference (FTC)

  • Home
  • Call for Papers
  • Submit your Paper/Poster
  • Register
  • Venue
  • Contact
  • Home
  • Call for Papers
  • Editorial Board
  • Guidelines
  • Submit
  • Current Issue
  • Archives
  • Indexing
  • Fees
  • Reviewers
  • Subscribe

IJACSA Volume 13 Issue 9

Copyright Statement: This is an open access publication licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, even commercially as long as the original work is properly cited.

View Full Issue

Paper 1: ModER: Graph-based Unsupervised Entity Resolution using Composite Modularity Optimization and Locality Sensitive Hashing

Abstract: Entity resolution describes techniques used to identify documents or records that might not be duplicated; nevertheless, they might refer to the same entity. Here we study the problem of unsupervised entity resolution. Current methods rely on human input by setting multiple thresholds prior to execution. Some methods also rely on computationally expensive similarity metrics and might not be practical for big data. Hence, we focus on providing a solution, namely ModER, capable of quickly identifying entity profiles in ambiguous datasets using a graph-based approach that does not require setting a matching threshold. Our framework exploits the transitivity property of approximate string matching across multiple documents or records. We build on our previous work in graph-based unsupervised entity resolution, namely the Data Washing Machine (DWM) and the Graph-based Data Washing Machine (GDWM). We provide an extensive evaluation of a synthetic data set. We also benchmark our proposed framework using state-of-the-art methods in unsupervised entity resolution. Furthermore, we discuss the implications of the results and how it contributes to the literature.

Author 1: Islam Akef Ebeid
Author 2: John R. Talburt
Author 3: Nicholas Kofi Akortia Hagan
Author 4: Md Abdus Salam Siddique

Keywords: Entity resolution; data curation; database; graph theory; natural language processing

PDF

Paper 2: Remote International Collaboration in Scientific Research Teams for Technology Development

Abstract: Scientific research teams often find themselves in remote working situations due to their internationality. Incredibly complex technological projects demand close collaboration and knowledge-sharing management. Remote teamwork, especially in cutting-edge scientific technology development, comes with various challenges that can negatively influence the overall team performance and commitment to the project. Within the EU-Japan (EU-/MIC-funded) project e-VITA, a consortium of 22 multidisciplinary partners and around 80 people work on research regarding a virtual assistant for healthy and active aging. We conducted qualitative data within the consortium after nine months of teamwork to understand the influence of collaboration on commitment, personal performance, efficiency, and work outcome. Based on this research's outcome, we built a framework for future scientific research projects and consortia to increase efficiency and quality of teamwork, thus researchers’ well-being.

Author 1: Sarah Janböcke
Author 2: Toshimi Ogawa
Author 3: Koki Kobayashi
Author 4: Ryan Browne
Author 5: Yasuki Taki
Author 6: Rainer Wieching
Author 7: Johanna Langendorf

Keywords: Teamwork; technology development; international collaboration; scientific team performance; potential technology leverage; scientific commitment; team efficiency; team commitment; team performance; collaboration

PDF

Paper 3: Fuzzy Image Enhancement Method based on a New Intensifier Operator

Abstract: In recent years, fuzzy image enhancement methods have been widely applied in image enhancement, which generally consists of three steps: fuzzification, modify membership(using intensifier (INT) operator), and defuzzification. This paper proposed a new INT operator used in fuzzy image enhancement. The INT operator is adjustable for different test images. The image enhancement method is as follows, firstly, calculate the image threshold (T ) using the OTSU method. Secondly, calculate pivotal point p corresponding to T , and find the corresponding INT operator function. Finally, use the INT operator in fuzzy Image Enhancement. The INT operator is used multiple times in the image processing process to obtain multiple result images. Comparative experiments show that the proposed new INT operator has better image enhancement effect when INT operator is applied at the same number of times. On the other hand, more intermediate process result images can also be obtained through the proposed new INT operator. More result images can provide material resources for the subsequent image processing.

Author 1: Libao Yang
Author 2: Suzelawati Zenian
Author 3: Rozaimi Zakaria

Keywords: Image enhancement; intensifier operator; threshold; pivotal point

PDF

Paper 4: Cooperative Multi-Robot Hierarchical Reinforcement Learning

Abstract: Recent advances in multi-robot deep reinforcement learning have made it possible to perform efficient exploration in problem space, but it remains a significant challenge in many complex domains. To alleviate this problem, a hierarchical approach has been designed in which agents can operate at many levels to complete tasks more efficiently. This paper proposes a novel technique called Multi-Agent Hierarchical Deep Deterministic Policy Gradient that combines the benefits of multiple robot systems with the hierarchical system used in Deep Reinforcement Learning. Here, agents acquire the ability to decompose a problem into simpler subproblems with varying time scales. Furthermore, this study develops a framework to formulate tasks into multiple levels. The upper levels function to learn policies for defining lower levels’ subgoals, whereas the lowest level depicts robot’s learning policies for primitive actions in the real environment. The proposed method is implemented and validated in a modified Multiple Particle Environment (MPE) scenario.

Author 1: Gembong Edhi Setyawan
Author 2: Pitoyo Hartono
Author 3: Hideyuki Sawada

Keywords: Multi-robot system; hierarchical deep reinforcement learning; path-finding; task decomposition

PDF

Paper 5: Differential Privacy Technology of Big Data Information Security based on ACA-DMLP

Abstract: Cloud computing and artificial intelligence have a deeper and closer connection with daily life. To ensure information security, most companies or individuals choose to pay a simple fee to store a large amount of data on cloud servers and hand over a large number of complex calculations of machine learning to cloud servers. To eliminate the security risks of data stored in the cloud and ensure that private data is not leaked, this paper proposes a collusion-resistant distributed machine learning scheme. Through homomorphic encryption algorithm and differential privacy algorithm, the security of data and model in machine learning framework is guaranteed. The distributed machine learning framework is adopted to reduce the data computing time and improve the data training efficiency. The simulation results show that the computational efficiency is improved while the user privacy security is guaranteed. The accuracy of model training is not reduced due to the improvement of privacy data security and computational efficiency. Through this study, we can further propose effective measures for the privacy protection of outsourced data and the data integrity of machine learning, which is of great significance to the security research of cloud intelligent big data.

Author 1: Yubiao Han
Author 2: Lei Wang
Author 3: Dianhong He

Keywords: Big data; cloud computing; information security; distributed machine learning; differential privacy algorithms

PDF

Paper 6: A Reusable Product Line Asset in Smart Mobile Application: A Systematic Literature Review

Abstract: A reusable product line asset is a product or asset that can be reused for different purposes including charity. Smart mobile applications are one of several communication and information methods used in charitable activities. Web, mobile, or hybrid platforms can be used to develop charity applications. It takes design and purpose to build an application, whether methodology or software development is applied for the smooth design or development of an application. The data for this study were acquired from the appropriate literature between 2017 and 2021 in order to determine the application development on current charity applications. The Systematic Literature Review (SLR) was employed in this study. The SLR method is used to identify, review, evaluate, and analyze all available research on relevant topics, as well as research issues for philanthropic development. This study aims to answer the following research questions: identify the donation applications that are frequently developed by researchers; identify the methods that are commonly used in the development of charity applications; identify the application platform that is frequently used; identify the functions utilized to the developed application and identify the key users who are using the application. The findings show, charity donations app, structured method, mobile applications, authentication and charity centers and donors were the most often observed in this study.

Author 1: Nan Pepin
Author 2: Abdul S. Shibghatullah
Author 3: Kasthuri Subaramaniam
Author 4: Rabatul Aduni Sulaiman
Author 5: Zuraida A. Abas
Author 6: Samer Sarsam

Keywords: Application; charity; donation; reusable product line; systematic literature review

PDF

Paper 7: A Study on the Effect of Digital Fabrication in Social Studies Education

Abstract: One of the learning methods that is increasingly being practiced in primary and secondary education is inquiry-based learning. This is not just a class to teach knowledge, but to practice activities to search for and discern the significance and essence of things. In social studies education, various trials and errors are being conducted, such as learning local history through fieldwork, and new approaches suitable for inquiry-based learning are being sought. In this study, as a new approach to social studies education, we developed a self-learning program that enables teachers to create original 3D educational materials using digital fabrication technology. We conducted an experiment in which students who wished to become social studies teachers participated in the program, created 3D educational materials, and taught a class using the materials. As a result, all the subjects who took the self-learning program could create 3D educational materials and give classes using them. The subjects' opinions suggested that practicing classes using 3D educational materials is effective for teacher education. This contributes to STEAM education, which has been spreading recently in the field of education, and this case study can be seen as a novel model.

Author 1: Kazunari Hirakoso
Author 2: Hidetake Hamada

Keywords: Digital fabrication; 3D educational materials; self-learning Program; social studies; STEAM education

PDF

Paper 8: A Blockchain-based Model for Securing IoT Transactions in a Healthcare Environment

Abstract: A blockchain is a data structure that is implemented as a distrusted database or digital ledger. The transactions are saved to a block of transactions that is attached in turn to the blockchain after the verification process, in which each block in the chain contains a hash signature of the previous block in addition to the hash signature of the block itself. The blocks on the blockchain are chained as an immutable list using the proof-of-work procedure, where there is no way to alter or delete an attached block due to the strict security policy used for structuring the chain of blocks. Each node holds a copy of the blockchain in which the miners take the responsibility of verifying and attaching blocks to the blockchain. The Ethereum blockchain introduced the smart contract which holds logic to be processed once the contract is established. These smart contracts are developed via the Solidity programming language. This proposed paper exploits the Ethereum blockchain along with smart contracts as the base technology for implementing the proposed blockchain-based model. The paper aims to develop a multilayered blockchain-based model, in which the blockchain model is set up on a private blockchain Ethereum network where the nodes share the electronic medical records (EMR) among the P2P (peer-to-peer) network that will be used to secure the IoT medical transactions. Solidity smart contract, introduced by Ethereum, is deployed to handle the EMR “open-query-transfer” operations on the private network, whereas the miners are responsible to validate the transactions. Finally, the research conducts a performance analysis of the Ethereum network using the Ethereum Caliper, considering several performance factors, which are: Maximum Latency, Minimum Latency, Average Latency, and Throughput.

Author 1: Mohamed Abdel Kader Mohamed Elgendy
Author 2: Mohamed Aborizka
Author 3: Ali Mohamed Nabil Allam

Keywords: Blockchain; ethereum; electronic medical records (EMR); ioi secure transactions; smart contracts; proof-of-work

PDF

Paper 9: Study on Early Warning on the Financial Risk of Project Venture Capital through a Neural Network Model

Abstract: This paper aims to effectively reduce the financial loss of enterprises by accurately and reasonably making early warning of investment project risks. This paper briefly introduced the index system used for investment project risk early warning. It constructed a project investment risk early-warning model with a back-propagation neural network (BPNN) algorithm, and improved it with a genetic algorithm (GA) to solve the defect that the traditional BPNN is easy to fall into, over-fitting when reversely adjust parameters. An analysis was conducted on an electric power company in Hunan Province. Orthogonal experiments are performed to determine the population size and the number of hidden layers in the improved BPNN algorithm. The results showed that the improved BPNN algorithm had the best performance when the population size was set as 25 and the number of hidden layers was four; compared with support vector machine (SVM) and traditional BPNN algorithms, the GA-improved BPNN algorithm had better performance for early risk warning of investment projects. In conclusion, adjusting the parameters of a BPNN with a GA in the training stage can effectively avoid falling into over-fitting, thus improving the early warning performance of the algorithm; in addition, the improved BPNN has better early warning performance.

Author 1: Xianjuan Li

Keywords: Neural network; project investment; early risk warning; genetic algorithm

PDF

Paper 10: Improving Privacy Preservation Approach for Healthcare Data using Frequency Distribution of Delicate Information

Abstract: In the modern world, everyone wishes that their personal information wouldn't be made public in any manner. In order to keep personal information hidden from prying eyes, privacy protection is essential. The data may be in the form of big data and minimization of risk and protection of sensitive data is important. In this research, a revolutionary customized privacy-preserving method is implemented that addresses the drawbacks of earlier personalized privacy as well as other anonymization methods. There are two main components that make up the proposed method's core. Delicate Information and Delicate Weight are two additional attributes which are used in the record table, are covered in the first section. The record holder's Delicate Information (DI) decides whether or not secrecy should be kept or if it should be shared. How delicate an attribute value is compared to the rest is indicated by its Delicate weight (DW). The second part covers a new representation used for anonymization termed the Frequency Distribution Block (FDB) and Quasi-Identifier Distribution Block (QIDB). According to experimental findings, the proposed system executes more quickly and with less data loss than current approaches.

Author 1: Ganesh Dagadu Puri
Author 2: D. Haritha

Keywords: Privacy preservation approach; quasi identifier distribution block; frequency distribution block; big data; anonymization

PDF

Paper 11: Attention-based Long Short Term Memory Model for DNA Damage Prediction in Mammalian Cells

Abstract: The understanding of DNA damage intensity – concentration-level is critical for biological and biomedical research, such as cellular homeostasis, tumor suppression, immunity, and gametogenesis. Therefore, recognizing and quantifying DNA damage intensity levels is a substantial issue, which requires further robust and effective approaches. DNA damage has several intensity levels. These levels of DNA damage in malignant cells and in other unhealthy cells are significant in the assessment of lesion stages located in normal cells. There is a need to get more insight from the available biological data to predict, explore and classify DNA damage intensity levels. Herein, the development process relied on the available biological dataset related to DNA damage signaling pathways, which plays a crucial role in DNA damage in the mammalian cell system. The biological dataset that was used in the proposed model consists of 15000 records intensity – concentration-level for a set of five proteins which regulate DNA damage. This research paper proposes an innovative deep learning model, which consists of an attention-based long short term-memory (AT-LSTM) model for DNA damage multi class predictions. The proposed model splits the prediction procedure into dual stages. For the first stage, we adopt the related feature sequences which are inserted as input to the LSTM neural network. In the next stage, the attention feature is applied efficiently to adopt the related feature sequences which are inserted as input to the softmax layer for prediction in the following frame. Our developed framework not only solves the long-term dependence problem of prediction effectively, but also enhances the interpretability of the prediction methods that was established on the neural network. We conducted a novel proposed model on big and complex biological datasets to perform prediction and multi classification tasks. Indeed, the (AT-LSTM) model has the ability to predict and classify the DNA damage in several classes: No-Damage, Low-damage, Medium-damage, High-damage, and Excess-damage. The experimental results show that our framework for DNA damage intensity level can be considered as state of the art for the biological DNA damage prediction domain.

Author 1: Mohammad A. Alsharaiah
Author 2: Laith H. Baniata
Author 3: Omar Adwan
Author 4: Ahmad Adel Abu-Shareha
Author 5: Mosleh Abu Alhaj
Author 6: Qasem Kharma
Author 7: Abdelrahman Hussein
Author 8: Orieb Abualghanam
Author 9: Nabeel Alassaf
Author 10: Mohammad Baniata

Keywords: Mammalian cell; deep learning techniques; attention; LSTM; classification; DNA damage

PDF

Paper 12: Estimation of Recovery Percentage in Gravimetric Concentration Processes using an Artificial Neural Network Model

Abstract: The concentrate process is the most sensitive in mineral processing plants (MPP), and the optimization of the process based on intelligent computational models (machine learning for recovery percentage modelling) can offer significant savings for the plant. Recent theoretical developments have revealed that many of the parameters commonly assumed as constants in gravity concentration modelling have a dynamic nature; however, there still lacks a universal way to model these factors accurately. This paper aims to understand the model effect of operational parameters of a jig (gravimetric concentrator) on the recovery percentage of the interest mineral (gold) through empirical modeling. The recovery percentage of mineral particles in a vibrated bed of big particles is studied by experimental data. The data used for the modelling were from experimental test in a pilot-scale jig supplemented by a two-month field sampling campaign for collecting 151 tests varying the most significant parameters (amplitude and frequency of pulsation, water flow, height of the artificial porous bed, and particle size). It is found the recovery percentage (%R) decreases with increasing pulsation amplitude (A) and frequency (F) when the size ratio of small to large particles (d/D) is smaller than 0.148. An empirical model was developed through machine learning techniques, specifically an artificial neural network (ANN) model was built and trained to predict the jig recovery percentage as a function of operation parameters and is then used to validate the recovery as a function of vibration conditions. The performance of the ANN model was compared with a new 65 experimental data of the recovery percentage. Results showed that the model (R2 = 0.9172 and RMSE = 0.105) was accurate and therefore could be efficiently applied to predict the recovery percentage in a jig device.

Author 1: Manuel Alejandro Ospina-Alarcón
Author 2: Ismael E. Rivera-M
Author 3: Gabriel Elías Chanchí-Golondrino

Keywords: Empirical modeling; dynamic gravimetric concentration model; gravimetric concentration; machine learning for recovery percentage modelling; mineral processing

PDF

Paper 13: Risk Prediction Applied to Global Software Development using Machine Learning Methods

Abstract: Software companies aim to develop high-quality software projects with the best global resources at the best cost. To achieve this global software development (GSD), an approach should be used which adopts work on projects across multiple distributed locations, and this is also known as distributed development. When companies attempt to implement GSD, they face numerous challenges owing to the nature of GSD and its differences from traditional methods. The objectives of this study were to identify the top software development factors that affect the overall success or failure of a software project using exploratory data analysis to find relationships between these factors, and to develop and compare risk prediction models that use machine learning classification techniques such as logistic regression, decision tree, random forest, support vector machine, K-nearest neighbors, and Naive Bayes. The findings of this study are as follows: in GSD, the top 18 factors influencing the software project are listed; and experiments show that the logistic regression and random forest models provide the best results, with an accuracy of 89% and 85%, respectively, and an area under the curve of 73% and 71%, respectively.

Author 1: Hossam Hassan
Author 2: Manal A. Abdel-Fattah
Author 3: Amr Ghoneim

Keywords: Global software development; distributed development; risk prediction model; machine learning

PDF

Paper 14: HelaNER 2.0: A Novel Deep Neural Model for Named Entity Boundary Detection

Abstract: Named entity recognition (NER) is a sequential labelling task in categorizing textual nuggets into specific types. Named entity boundary detection can be recognized as a prominent research area under the NER domain which has been heavily adapted for information extraction, event extraction, information retrieval, sentiment analysis etc. Named entities (NE) can be identified as per flat NEs and nested NEs in nature and limited research attempts have been made for nested NE boundary detection. NER in low resource settings has been identified as a current trend. This research work has been scoped down to unveil the uniqueness in NE boundary detection based on Sinhala related contents which have been extracted from social media. The prime objective of this research attempt is to enhance the approach of named entity boundary detection. Considering the low resource settings, as the initial step, the linguistic patterns, complexity matrices and structures of the extracted social media statements have been analyzed further. A dedicated corpus of more than 100,000 tuples of Sinhala related social media content has been annotated by an expert panel. As per the scientific novelties, NE head word detection loss function, which was introduced in HelaNER 1.0, has been further improved and the NE boundary detection has been further enhanced through tuning up the stack pointer networks. Additionally, NE linking has been improved as a by-product of the previously mentioned enhancements. Various experimentations have been conducted, evaluated and the outcome has revealed that our enhancements have achieved the state-of-art performance over the existing baselines.

Author 1: Y. H. P. P Priyadarshana
Author 2: L Ranathunga

Keywords: Computational linguistics; deep neural networks; natural language processing; named entity boundary detection; named entity recognition

PDF

Paper 15: Face Recognition under Illumination based on Optimized Neural Network

Abstract: Face recognition is a significant area of pattern recognition and computer vision research. Illumination in face recognition is obvious yet challenging task in pattern matching. Recent researchers introduced machine learning algorithms to solve illumination problems in both indoor and outdoor scenarios. The major challenge in machine learning is the lack of classification accuracy. Thus, the novel Optimized Neural Network Algorithm (ONNA) is used to solve the aforementioned drawback. First, we propose a novel Weight Transfer Ideal Filter (WTIF) which is employed for pre-processing to remove the dark spots and shadows in an image by normalizing low frequency and high frequency of illumination. Secondly, Robust Principal Component Analysis (RPCA) is employed to perform efficient extraction of features based on image area representation. These features are given as input to ONNA which classifies the given input image under illumination. Thus we achieve the recognition of the face under various illumination conditions. Our approach is analyzed and compared with existing approaches such as Support Vector Machine (SVM) and Random Forest (RF). ONNA is better in terms of high accuracy and low error rate.

Author 1: Napa Lakshmi
Author 2: Megha P Arakeri

Keywords: Face recognition; illumination; neural network; robust principal component analysis

PDF

Paper 16: Transfer Learning for Medicinal Plant Leaves Recognition: A Comparison with and without a Fine-Tuning Strategy

Abstract: Plant leaves are another common source of information for determining plant species. According to the dataset that has been collected, we propose transfer learning models VGG16, VGG19, and MobileNetV2 to examine the distinguishing features to identify medicinal plant leaves. We also improved algorithm using fine-tuning strategy and analyzed a comparison with and without a fine-tuning strategy to transfer learning models performance. Several protocols or steps were used to conduct this study, including data collection, data preparation, feature extraction, classification, and evaluation. The distribution of training and validation data is 80% for training data and 20% for validation data, with 1500 images of thirty species. The testing data consisted of a total of 43 images of 30 species. Each species class consists of 1-3 images. With a validation accuracy of 96.02 percent, MobileNetV2 with fine-tuning had the best validation accuracy. MobileNetV2 with fine-tuning also had the best testing accuracy of 81.82%.

Author 1: Vina Ayumi
Author 2: Ermatita Ermatita
Author 3: Abdiansah Abdiansah
Author 4: Handrie Noprisson
Author 5: Yuwan Jumaryadi
Author 6: Mariana Purba
Author 7: Marissa Utami
Author 8: Erwin Dwika Putra

Keywords: Medicinal leaf plant; transfer learning; deep learning; phytomedicine

PDF

Paper 17: Effect of Random Splitting and Cross Validation for Indonesian Opinion Mining using Machine Learning Approach

Abstract: Opinion mining has been a prominent topic of research in Indonesia, however there are still many unanswered questions. The majority of past research has been on machine learning methods and models. A comparison of the effects of random splitting and cross-validation on processing performance is required. Text data is in Indonesian. The goal of this project is to use a machine learning model to conduct opinion mining on Indonesian text data using a random splitting and cross validation approach. This research consists of five stages: data collection, pre-processing, feature extraction, training & testing, and evaluation. Based on the experimental results, the TF-IDF feature is better than the Count-Vectorizer (CV) for Indonesian text. The best accuracy results are obtained by using TF-IDF as a feature and Support Vector Machine (SVM) as a classifier with cross validation implementation. The best accuracy reaches 81%. From the experimental results, it can also be seen that the implementation of cross validation can improve accuracy compared to the implementation of random splitting.

Author 1: Mariana Purba
Author 2: Ermatita Ermatita
Author 3: Abdiansah Abdiansah
Author 4: Handrie Noprisson
Author 5: Vina Ayumi
Author 6: Hadiguna Setiawan
Author 7: Umniy Salamah
Author 8: Yadi Yadi

Keywords: Random splitting; cross validation; machine learning; Indonesian text

PDF

Paper 18: Classification of Diabetes Types using Machine Learning

Abstract: Machine learning algorithms have aided health workers (including doctors) in the processing, analysis, and diagnosis of medical problems, as well as the detection of disease patterns and other patient data. Diabetes mellitus (DM), commonly referred to as diabetes, is a gathering of a syndrome issue that is portrayed by high glucose levels in the blood over a drawn-out period. It is a long-term illness that is a great threat to humanity and causes death. Most of the existing machine learning algorithms used for the classification and prediction of diabetes suffer from embodying redundant or inessential medical procedures that cause complications and wastage of time and resources. The absence of a correct diagnosis scheme, deficiency of economic means, and a general lack of awareness represent the main reasons for these negative effects. Hence, preventing the sickness altogether through early detection may doubtless cut back a considerable burden on the economy and aid the patient in diabetes management. This study developed diabetes classification using machine learning techniques that will minimize the aforementioned drawbacks in the prediction of diabetes systems. Decision tree classifiers, logistic regression, random forest, and support vector machines are all examples of predictive algorithms that were tested in this paper. 1009 records of data set were obtained from the Diabetes dataset of Abelvikas, Data World. We used a confusion matrix to visualize the performance evaluation of the classifiers. The experimental result shows that the four machine learning algorithms perform well. However, Random Forest outperforms the other three, with a prediction accuracy of 100% and has a better prediction level when compared with others and existing work.

Author 1: Oyeranmi Adigun
Author 2: Folasade Okikiola
Author 3: Nureni Yekini
Author 4: Ronke Babatunde

Keywords: Machine learning; diabetes mellitus; predictive algorithm; correlation map; confusion matrix

PDF

Paper 19: Criteria and Guideline for Dyslexic Intervention Games

Abstract: The utilization of game-based interventions is growing as a result of technological advancements, and it has shown to be effective in the treatment of dyslexia and other medical conditions. Games are typically viewed as activities having the essential components of challenge, incentive, and reward. Games were originally created for pleasure, and they can make dyslexic teaching and learning more enjoyable and exciting. Although there are numerous applications available for treating dyslexic children, the inclusion of games and their standards in those applications has not yet been established. Therefore, there is a need for a standard design guideline to be formulated in establishing a guideline for designing and developing games specifically for dyslexic children. This article proposes a design guideline for dyslexic intervention games. Two methods have been employed which are interviews and systematic literature reviews (SLR) to discover the characteristics of dyslexic games. The first set of the criteria was developed through interviews with the stakeholders who are directly associated with dyslexic children. Scopus, the ACM digital library, EBSCO-host, Wiley, and Web of Science (WOS) are the five primary databases used in SLR. 50 articles out of the 551 that were early screened from the five primary databases are qualified to be studied based on the criteria. Only 23 publications could be selected for the study after further screening, which led to the creation of a second set of criteria. These two sets of criteria are thoroughly analyzed, combined, and formulated as a guideline which comprises of four main categories; device and platform, interface, game features, and gameplay. The guideline consists of guidance to be used for designing and developing Dyslexic therapy games with the purpose of assisting Dyslexic children to read. The guideline is believed to be beneficial to many parties especially the educational game developers, therapists, and educationist who are dealing with intervention for Dyslexic children. This study is aligned and significant to Sustainable Development Goals (SDG) three and four, Good Health and Well-being and Quality Education respectively.

Author 1: Noraziah ChePa
Author 2: Nur Azzah Abu Bakar
Author 3: Laura Lim Sie-Yi

Keywords: Dyslexic therapy games; game-based intervention; specific learning disorder; guideline for dyslexia games; dyslexia intervention

PDF

Paper 20: The Effectiveness of Gamification for Students' Engagement in Technical and Vocational Education and Training

Abstract: The transformation of Technical and Vocational Education and Training (TVET) prioritizes by the national education convention to meet the needs of the industry through improving student skills and the quality of related systems. One of the transformations is practicing blended learning, such as a flipped classroom, to produce better quality student learning outcomes. However, based on previous studies, there are difficulties in maintaining student engagement during learning activities, even though blended learning offers some advantages. Therefore, this study suggests the development of a mobile application using gamification as a solution to enhance student participation. This paper proposes the design and development research (DDR) approach with the adaptation of the ADDIE model to build a learning content prototype. It involves five phases: analysis, design, development, implementation, and evaluation. The study participants consisted of two groups of students in the 1st semester of the Interactive Multimedia course from two different TVET institutions who were cleft into a control group and an experimental group. The experimental group is gamified, whereas the control group is not. The study evaluation uses two instruments: a test to compare students' understanding of both groups and an activity log to track the experimental group's use of the prototype. According to the findings, gamification during learning activities can increase student engagement by boosting performance through a more significant pre-and post-test mean score difference and creating a positive learning experience. Additionally, mobile applications with the gamification concept can be employed extensively in various TVET courses to encourage student learning performance.

Author 1: Laily Abu Samah
Author 2: Amirah Ismail
Author 3: Mohammad Kamrul Hasan

Keywords: Technical and vocational education and training (TVET); flipped classroom; engagement; gamification; mobile application

PDF

Paper 21: Campus Quality of Services Analysis of Mobile Wireless Communications Network Signal among Providers in Malaysia

Abstract: Wireless communication is very important in this generation where todays 5G internet connection is still unconfirmed and 4G communication is still needed. Network in Malaysia has been supported by many telecommunication companies and the Quality of Services is still poor supported especially in the campus area. This research presents a performance analysis of Quality of Services for 4G wireless Communication among Providers supported in a campus area in Malaysia. A 4G Nemo Outdoor wireless analyzer was used to collect the Reference Signal Received Power (RSRP) signal data based on the identified campus road maps. Digi and U-Mobile Network was identified and compared as two telecommunications providers in the testing. The identified road maps were analyzed along the routes while testing signals are collected while driving. It is identified that Digi supports better for the Mobile broadband network which shows an excellent of 1% and good connections of 29 % and 0% signal loss in the drive areas. RSRP signal for U-Mobile shows there is 8% signal loss and the connections provided only at the Mid-Cell for 43% and Cell Edge connections for 48%. This concludes that the 4G signal strength in the campus area having average signal strength, but some medium signal strength is also identified based on the road locations. This research is significant for QoS of supports mobile network in a campus area.

Author 1: Murizah Kassim
Author 2: Zulfadhli Hisam
Author 3: Mohd Nazri Ismail

Keywords: Quality of services; 4G/LTE; mobile network; wireless communication; RSRP; campus network

PDF

Paper 22: Exploring Alumni Data using Data Visualization Techniques

Abstract: Alumni data are mostly managed through paper-based and word file. With lots of alumni graduating each year, these massive data become difficult to handle. It is hard to look for past alumni data to know their current situation. Since the data is kept conventionally, there is no communication between the alumni and the faculty. Therefore, we proposed a solution that includes alumni information regarding their status in life where alumni themselves and individuals in the faculty can see. This study aims to visualize alumni data from a faculty in a public university through an exploratory dashboard using the identified data visualization techniques. This study adopts the dashboard development process consists of three major phases, which are conception, visualization, and finalization phase. The primary audience is identified and the theme for the dashboard is decided in the conceptual phase. The primary and support views are then designed together with the layout during the visualization phase. At the end of the study, the exploratory dashboard for alumni data using multidimensional and hierarchical data visualization is finalized with the interactive elements. The results are interpreted through descriptive and diagnostic analysis. The dashboard is then evaluated through convenience sampling technique to verify the representation of the dashboard. Majority respondents agreed on the simplicity of exploratory dashboard and the amount of data is also sufficient with the selection of the visualization types. The dashboard is beneficial to the university’s administrator, alumni, and public.

Author 1: Nurhanani Izzati Ismail
Author 2: Nur Atiqah Sia Abdullah
Author 3: Nasiroh Omar

Keywords: Alumni; descriptive analysis; diagnostic analysis; data visualization; exploratory dashboard

PDF

Paper 23: The Performance Evaluation of Transfer Learning VGG16 Algorithm on Various Chest X-ray Imaging Datasets for COVID-19 Classification

Abstract: Early detection of the coronavirus (COVID-19) disease is essential in order to contain the spread of the virus and provide effective treatment. Chest X-rays could be used to detect COVID-19 at an early stage. However, the pathological features of COVID-19 on chest X-rays closely resemble those caused by other viruses. The visual geometry group-16 (VGG16) deep learning algorithm based on convolutional neural network (CNN) architecture is commonly used to detect various pathologies on medical images automatically and may have a role in the detection of COVID-19 on chest X-rays. Therefore, this research is aimed to determine the robustness of the VGG16 architecture on several chest X-ray databases that vary in terms of size and the number of class labels. Nine publicly available chest X-ray datasets were used to train and test the algorithm. Each dataset had a different number of images, class compositions, and interclass proportions. The performance of the architecture was tested using several scenarios, including datasets above and below 5,000 samples, label class variation, and interclass ratio. This study confirmed that the VGG16 delivers robust performance on various datasets, achieving an accuracy of up to 97.99%. However, our findings also suggest that the accuracy of the VGG16 algorithm drops drastically in highly imbalanced datasets.

Author 1: Andi Sunyoto
Author 2: Yoga Pristyanto
Author 3: Arief Setyanto
Author 4: Fawaz Alarfaj
Author 5: Naif Almusallam
Author 6: Mohammed Alreshoodi

Keywords: Covid-19; Chest X-Ray; CNN; transfer learning; VGG-16

PDF

Paper 24: A Comprehensive Review and Application of Interpretable Deep Learning Model for ADR Prediction

Abstract: Drug safety is a pressing need in today's healthcare. Minimizing drug toxicity and improving the individual’s health and society is the key objective of the healthcare domain. Drugs are clinically tested in laboratories before marketing as medicines. However, the unintended and harmful effects of drugs are called Adverse Drug Reactions (ADRs). The impact of ADRs can range from mild discomfort to more severe health hazards leading to hospitalization and in some cases death. Therefore, the objective of this research paper is to design a framework based on which research papers are collected from both ADR detection and prediction domain. Around 172 research articles are collected from the sites like ResearchGate, PubMed, etc. After applying the elimination criteria the author categorized them into ADR detection and prediction themes. Further, common data sources and algorithms as well as the evaluation metrics were analyzed and their contribution to their respective domains is stated in terms of percentages. A deep learning framework is also designed and implemented based on the research gaps identified in the existing ADR detection and prediction models. The performance of the deep learning model with two hidden layers was found to be optimum for ADR prediction and further, the non-interpretability part of the model is addressed using a global surrogate model. The proposed architecture has successfully addressed multiple limitations of existing models and also highlights the importance of early detection & prediction of adverse drug reactions in the healthcare industry.

Author 1: Shiksha Alok Dubey
Author 2: Anala A. Pandit

Keywords: Drug safety; adverse drug reactions; early detection; deep learning; interpretable models

PDF

Paper 25: Secure Cloud Connected Indoor Hydroponic System via Multi-factor Authentication

Abstract: Now-a-days, the hydroponic farming system with the Internet of Things (IoT) technology is increasingly becoming a trend for researchers to produce a more capable farming device or remote monitoring system. However, this intelligent system is not controlled securely and will be dangerous when system hacking occurs. Therefore, developing a secure indoor hydroponic monitoring device with multi-factor authentication (MFA) method is proposed. The research aims to develop a secure cloud-connected indoor hydroponic system via multi-factor authentication on the ThingsSentral IoT platform with an MFA technique. The developed system comprises an iPhone Operating System (iOS), an Arduino node microcontroller unit and a ThingsSentral web IoT platform. A security software application on iOS phones with MFA techniques is built to authenticate devices before communicating with ThingsSentral.io. Token authentication between ThingsSentral.io and the security software application must be done before the hydroponic monitoring device can send and receive data. An indoor hydroponic monitoring system device with MFA security technique has been successfully produced from the study. An MFA security technique for iOS apps has also been successfully developed. In conclusion, using the MFA technique, this research successfully develops a high-security control and communication system between the field device and the IoT platform. Although the MFA security system developed for this IoT platform has several steps that need to be done before data can be sent to the cloud database, the users themselves can allow or prohibit a device from operating. Besides, users can also control and monitor the security between the device and the IoT platform when they operate.

Author 1: Mohamad Khairul Hafizi Rahimi
Author 2: Mohamad Hanif Md Saad
Author 3: Aini Hussain
Author 4: Nurul Maisarah Hamdan

Keywords: Internet of things; intelligent system; remote monitoring; hydroponic; multi-factor authentication

PDF

Paper 26: Effective Multitier Network Model for MRI Brain Disease Prediction using Learning Approaches

Abstract: Brain disease prognosis is considered a hot research topic where the researchers intend to predict the clinical measures of individuals using MRI data to evaluate the pathological stage and identifies the progression of the disease. With the lack of incomplete clinical scores, various existing learning-based approaches simply eradicate the score without ground truth score computation. It helps restrict the training data samples with robust and reliable models during the learning process. The major disadvantage of the prior approaches is the adoption of hand-crafted features, as these features are not well-suited for the prediction process. This research concentrates on modelling a weakly supervised multi-tier dense neural network model (ws-MTDNN) for examining the progression of brain disease using the available MRI data. The model helps analyze the incomplete clinical scores. The preliminary ties of the network model initially haul out the distinctive patches from the MRI to extract the global and local structural features (information) and develop a superior multi-tier dense neural network model for task-based image feature extraction and perform prediction in the successive tiers for computing the clinical measures. The loss function is adopted while examining the available individuals even in the absence of ground-truth values. The experimentation is done with the available online Dataset like ADNI-1/2, and the model works effectually with this Dataset compared to other approaches.

Author 1: N. Ravinder
Author 2: Moulana Mohammed

Keywords: Brain disease; learning approaches; ground truth value; feature learning; global and local feature analysis

PDF

Paper 27: Application based on Hybrid CNN-SVM and PCA-SVM Approaches for Classification of Cocoa Beans

Abstract: In our study, we propose a hybrid Convolutional Neural Network with Support Vector Machine (CNN-SVM) and Principal Component Analysis with support vector machine (PCA-SVM) methods for the classification of cocoa beans obtained by the fermentation of beans collected from cocoa pods after harvest. We also use a convolutional neural network (CNN) and support vector machine (SVM) for the classification operation. In the case of the hybrid model, we use a convolutional network as a feature extractor and the SVM is used to perform the classification operation. The use of PCA-SVM allowed for a reduction in image size while maintaining the main features still using the SVM classifier. Radial, linear and polynomial basis function kernels were used with various control parameters for the SVM, and optimizers such as the Stochastic Gradient Descent (SGD) algorithm, Adam, and RMSprop were used for the CNN softmax classifier. The results showed the robustness of the hybrid CNN-SVM model which obtained the best score with a value of 98.32% then the PCA-SVM based model had a score of 97.65% outperforming the standard CNN and SVM classification algorithms. Metrics such as accuracy, recall, F1 score, mean squared error (MSE), and MCC have allowed us to consolidate the results obtained from our different experiments.

Author 1: AYIKPA Kacoutchy Jean
Author 2: MAMADOU Diarra
Author 3: BALLO Abou Bakary
Author 4: GOUTON Pierre
Author 5: ADOU Kablan Jérôme

Keywords: Support vector machine; convolutional neural network; cocoa beans; principal component analysis; hybrid method

PDF

Paper 28: SQrum: An Improved Method of Scrum

Abstract: Software systems are having a major impact on many aspects of personal and professional life. Safety-critical applications, such as production line controls, automotive operations, and process industry controls, rely significantly on software systems. In these applications, software failure may result in bodily damage or death. The proper operation of software is essential to the safety and well-being of individuals and businesses. Therefore, software quality assurance is of paramount relevance in the software business today. In recent years, Agile Project Management and particularly Scrum, have gained popularity as a method of dealing with "vuca" business environments, which are characterized by rising Volatility, Uncertainty, Complexity, and Ambiguity. This paper contributes to the software development body of knowledge by proposing a metamodel of Scrum quality assurance, named SQrum (‘SQ’ of Software Quality and ‘rum’ of Scrum). Our objective is to make Scrum more efficient and reliable and to assist enterprises in undertaking quality assurance activities while considering agile practices and values.

Author 1: Najihi Soukaina
Author 2: Merzouk Soukaina
Author 3: Marzak Abdelaziz

Keywords: Agile project management; IT; OMG; meta-object facility; MOF; metamodel; scrum; SQrum; quality assurance; QA; quality management; QM; software development project

PDF

Paper 29: Use of Interactive Multimedia e-Learning in TVET Education

Abstract: Malaysia is focused on the development and use of technologies among consumers. Thus, technological innovations are used in the adaptation of online learning to educate students, as well as to enhance the teaching and learning process in Technical and Vocational Education and Training (TVET) institutions. There is a need to expose students to the online learning revolution, which conceptualises using computerised systems to facilitate the learning process. However, the COVID-19 outbreak has disrupted the academic year across the country. Due to the unusual circumstances related to the pandemic, the Malaysian government has urged all academic institutions to conduct online teaching and learning. Thus, an e-Learning system, known as SpmiILP, has been designed and developed accordingly for an interactive multimedia course to encourage online interaction among students and lecturers, as well as to enhance human learning and cognitive development. In fact, these essential elements such as learning style of the students and user experience are focused on to engage them in learning effectively as well. An e-Learning System for Interactive Multimedia Course was used to develop the e-Learning system (SpmiILP). The usability test showed that the developed e-Learning system has a positive influence that provided potential contributions to (TVET) students in their learning processes.

Author 1: Siti Fadzilah Mat Noor
Author 2: Hazura Mohamed
Author 3: Nur Atiqah Zaini
Author 4: Dayana Daiman

Keywords: e-Learning; interactive multimedia; learning style; user experience

PDF

Paper 30: CBT4Depression: A Cognitive Behaviour Therapy (CBT) Therapeutic Game to Reduce Depression Level among Adolescents

Abstract: Dropping out of depression treatment commonly occurs in the current psychotherapy treatment. Adolescents often find it difficult to express their thoughts and feelings clearly due to their developmental constraints. They also have trouble realising their behaviours as unhealthy or problematic. The use of therapeutic games in depression treatment among adolescents can enhance the engagement level. Indirectly, the issue of dropping out can be reduced among the adolescents. Therefore, this study aimed to improve engagement levels and reduce depression level among adolescents with depression by designing a therapeutic game. A prototype named CBT4Depression was developed in this study. A quasi experimental study was conducted to evaluate the developed therapeutic game and 115 adolescents were recruited to measure their depression level using CBT4Depression. Based on the findings from the evaluation process, it can be concluded that the CBT4Depression considered success to engage and reduce the depression level among adolescents.

Author 1: Norhana Yusof
Author 2: Nazrul Azha Mohamed Shaari
Author 3: Eizwan Hamdie Yusoff

Keywords: Therapeutic; game; depression; adolescents; cognitive behavior therapy

PDF

Paper 31: Creating Video Visual Storyboard with Static Video Summarization using Fractional Energy of Orthogonal Transforms

Abstract: The overwhelming number of video uploads and downloads has made it incredibly difficult to find, gather, and archive videos. A static video summarization technique highlights an original video's significant points through a set of static keyframes as a video visual storyboard. The video visual storyboards are created as static video summaries that solve video processing-related issues like storage and retrieval. In this paper, a strategy for effectively summarizing static videos using the feature vectors, which are fractional coefficients of the transformed video frames, is proposed and evaluated. Four popular orthogonal transforms are deployed for generating feature vectors of video frames. The fractional coefficients of transformed video frames taken as 25 percent, 6.25 percent, and 1.5625 percent of full 100 percent transformed coefficients are considered to form video visual storyboards. The proposed method uses the benchmark video datasets Open Video Project (OVP) and SumMe to validate the performance, containing user summaries (storyboards). These video summaries created using the proposed method are evaluated using percentage accuracy and matching rate.

Author 1: Ashvini Tonge
Author 2: Sudeep D. Thepade

Keywords: Keyframe; orthogonal transform; VSUMM; video visual storyboard; video summarization

PDF

Paper 32: Denoising of Impulse Noise using Partition-Supported Median, Interpolation and DWT in Dental X-Ray Images

Abstract: The impulse noise often damages the human dental X-Ray images, leading to improper dental diagnosis. Hence, impulse noise removal in dental images is essential for a better subjective evaluation of human teeth. The existing denoising methods suffer from less restoration performance and less capacity to handle massive noise levels. This method suggests a novel denoising scheme called "Noise removal using Partition supported Median, Interpolation, and Discrete Wavelet Transform (NRPMID)" to address these issues. To effectively reduce the salt and pepper noise up to a range of 98.3 percent noise corruption, this method is applied over the surface of dental X-ray images based on techniques like mean filter, median filter, Bi-linear interpolation, Bi-Cubic interpolation, Lanczos interpolation, and Discrete Wavelet Transform (DWT). In terms of PSNR, IEF, and other metrics, the proposed noise removal algorithm greatly enhances the quality of dental X-ray images.

Author 1: Mohamed Shajahan
Author 2: Siti Armiza Mohd Aris
Author 3: Sahnius Usman
Author 4: Norliza Mohd Noor

Keywords: Salt and pepper noise; impulse noise; X-ray noise removal; X-ray teeth image quality enhancement; dental X-ray noise reduction

PDF

Paper 33: An End-to-End Big Data Deduplication Framework based on Online Continuous Learning

Abstract: While big data benefits are numerous, most of the collected data is of poor quality and, therefore, cannot be effectively used as it is. One of the leading big data quality challenges is data duplication. Indeed, the gathered big data are usually messy and may contain duplicated records. The process of detecting and eliminating duplicated records is known as Deduplication, or Entity Resolution or also Record Linkage. Data deduplication has been widely discussed in the literature, and multiple deduplication approaches were suggested. However, few efforts have been made to address deduplication issues in Big Data Context. Also, the existing big data deduplication approaches are not handling the case of the decreasing performance of the deduplication model during the serving. In addition, most current methods are limited to duplicate detection, which is part of the deduplication process. Therefore, we aim through this paper to propose an End-to-End Big Data Deduplication Framework based on a semi-supervised learning approach that outperforms the existing big data deduplication approaches with an F-score of 98,21%, a Precision of 98,24% and a Recall of 96,48%. Moreover, the suggested framework encompasses all data deduplication phases, including data preprocessing and preparation, automated data labeling, duplicate detection, data cleaning, and an auditing and monitoring phase. This last phase is based on an online continual learning strategy for big data deduplication that allows addressing the decreasing performance of the deduplication model during the serving. The obtained results have shown that the suggested continual learning strategy has increased the model accuracy by 1,16%. Furthermore, we apply the proposed framework to three different datasets and compare its performance against the existing deduplication models. Finally, the results are discussed, conclusions are made, and future work directions are highlighted.

Author 1: Widad Elouataoui
Author 2: Imane El Alaoui
Author 3: Saida El Mendili
Author 4: Youssef Gahi

Keywords: Big data deduplication; online continual learning; big data; entity resolution; record linkage; duplicates detection

PDF

Paper 34: Student’s Performance Prediction based on Personality Traits and Intelligence Quotient using Machine Learning

Abstract: Apparently, most life activities that people perform depend on their unique characteristics. Personal characteristics vary across people, so they perform tasks in different ways based on their skills. People have different mental, psychological, and behavioral features that affect most life activities. This is the same case with students at various educational levels. Students have different features that affect their academic performance. The academic score is the main indicator of the student’s performance. However, other factors such as personality features, intelligence level, and basic personal data can have a great influence on the student’s performance. This means that the academic score is not the only indicator that can be used in predicting students’ performance. Consequently, an approach based on personal data, personality features, and intelligence quotient is proposed to predict the performance of university undergraduates. Five machine learning techniques were used in the proposed approach. In order to evaluate the performance of the proposed approach, a real student’s dataset was used, and various performance measures were computed. Several experiments were performed to determine the impact of various features on the student’s performance. The proposed approach gave promising results when tested on the dataset.

Author 1: Samar El-Keiey
Author 2: Dina ElMenshawy
Author 3: Ehab Hassanein

Keywords: Prediction; student performance; machine learning; personality; intelligence quotient

PDF

Paper 35: Real Time Fire Detection using Color Probability Segmentation and DenseNet Model for Classifier

Abstract: The forest is an outdoor environment not touched by the surrounding community, so it is not immediately handled when a fire occurs. Therefore, surveillance using cameras is needed to see the presence of fire hotspots in the forest. This study aims to detect hotspots through video data. As is known, fire has a variety of colors, ranging from yellow to reddish. The segmentation process requires a method that can recognize various fire colors to get a candidate fire object area in the video frame. The methods used for the color segmentation process are Gaussian Mixture Model (GMM) and Expectation–maximization (EM). The segmentation results are candidates for fire areas, which in the experiment used the value of K=4. This fire object candidate needs to be ascertained whether the segmented object is a fire object or another object. In the feature extraction stage, this research uses the DenseNet-169 or DenseNet-201 models. In this study, various color tests were carried out, namely RGB, HSV, and YCbCr. The test results show that RGB color produces the most optimal training accuracy. This RGB color configuration is used to test using video data. The test results show that the true positive and false negative values are quite good, 98.69% and 1.305%. This video data processing produces fps with an average of 14.43. So, it can be said that this combination of methods can be used to process real time data in case studies of fire detection.

Author 1: Faisal Dharma Adhinata
Author 2: Nur Ghaniaviyanto Ramadhan

Keywords: Fire detection; color segmentation; GMM-EM; DenseNet; real time

PDF

Paper 36: Tissue and Tumor Epithelium Classification using Fine-tuned Deep CNN Models

Abstract: The field of Digital Pathology (DP) has become more interested in automated tissue phenotyping in recent years. Tissue phenotyping may be used to identify colorectal cancer (CRC) and distinguish various cancer types. The information needed to construct automated tissue phenotyping systems has been made available by the introduction of Whole Slide Images (WSIs). One of the typical pathological diagnosis duties for pathologists is the histopathological categorization of epithelial tumors. Artificial intelligence (AI) based computational pathology approaches would be extremely helpful in reducing the pathologists ever-increasing workloads, particularly in areas where access to pathological diagnosis services is limited. Investigating several deep learning models for categorizing the images of tumor epithelium from histology is the initial goal. The varying accuracy ratings that were achieved for the deep learning models on the same database demonstrated that additional elements like pre-processing, data augmentation, and transfer learning techniques might affect the models' capacity to attain better accuracy. The second goal of this publication is to reduce the time taken to classify the tissue and tumor Epithelium. The final goal is to examine and fine-tune the most recent models that have received little to no attention in earlier research. These models were checked by the histology Kather CRC image database's nine classifications (CRC-VAL-HE-7K, NCT-CRC-HE-100K). To identify and recommend the most cutting-edge models for each categorization, these models were contrasted with those from earlier research. The performance and the achievements of the proposed preprocessing workflow and fine-tuned Deep CNN models (Alexnet, GoogLeNet and Inceptionv3) are greater compared to the prevalent methods.

Author 1: Anju T E
Author 2: S. Vimala

Keywords: Colorectal cancer; deep learning; CNN; tumor epithelium; Alexnet; GoogLeNet; Inceptionv3

PDF

Paper 37: Predicting University Student Retention using Artificial Intelligence

Abstract: Based on the advancement in the field of Artificial Intelligence, there is still a room for enhancement of student university retention. The main objective of this study is to assess the probability of using Artificial Intelligence techniques such as deep and machine learning procedures to predict university student retention. In this study a variable assessment is carried out on the dataset which was collected from Kaggle repository. The performance of twenty supervised algorithms of machine learning and one algorithm of deep learning is assessed. All algorithms were trained using 10 variables from 1100 records of former university student registrations that have been registered in the University. The top performing algorithm after hyper-parameters tuning was NuSVC Classifier. Therefore, we were able to use the current dataset to create supervised Machine Learning (ML) and Deep Learning (DL) models for predicting student retention with F1-score (90.32 percent) for ML and the proposed DL algorithm with F1-score (93.05 percent).

Author 1: Samer M. Arqawi
Author 2: Eman Akef Zitawi
Author 3: Anees Husni Rabaya
Author 4: Basem S. Abunasser
Author 5: Samy S. Abu-Naser

Keywords: Artificial intelligence; machine learning; deep learning; retention; student; prediction

PDF

Paper 38: Using the Agglomerative Hierarchical Clustering Method to Examine Human Factors in Indonesian Aviation Accidents

Abstract: This study aims to provide a comprehensive source of knowledge regarding aviation accidents in Indonesia caused by human factors, which are the most significant among other causative elements, requiring a detailed assessment of the accident as a result of pilot and co-pilot faults while operating the aircraft. The KNKT website database is still in the form of accident reports. To this end, the retrieved information based on historical data for 23 years of accidents caused by humans by analyzing the data using the clustering approach to gain data insight in the relationship between total flying hours and pilot licenses. The data analysis revealed that, in general, the aircraft operator complied with the CASR standards.

Author 1: Rossi Passarella
Author 2: Gulfi Oktariani
Author 3: Dedy Kurniawan
Author 4: Purwita Sari

Keywords: Aviation accidents data; pilot’s licenses; flying hours; human factor

PDF

Paper 39: A Framework for Crime Detection and Diminution in Digital Forensics (CD3F)

Abstract: Cyber-attacks have become one of the world's most serious issues. Every day, they wreak serious financial harm to governments and people. As cyber-attacks become more common, so does cyber-crime. Identifying cyber-crime perpetrators and understanding attack tactics are critical in the battle against crime and criminals. Cyber-attack detection and prevention are difficult undertakings. Researchers have lately developed security models and made forecasts using artificial intelligence technologies to solve these concerns. In the literature, the authors explained numerous ways of predicting crime. They, on the other hand, have a problem forecasting cyber-crime and cyber-attack strategies. Here, in this paper author proposed a digital forensic investigation procedure that deals with cyber-crime. In this investigation, the process author explains digital forensics techniques for ensuring that digital evidence is located, collected, preserved, evaluated, and reported in such a way that the evidence's integrity is preserved. These sequential digital forensic stages affect a standard and accepted digital forensic investigation procedure, and each phase is influenced by sequential occurrences, with each event relying on tasks. Digital forensics investigation is a technique for ensuring that digital evidence is handled in such a way that the evidence's integrity is preserved. Sequential digital forensic stages affect a standard and accepted digital forensic investigation procedure, and each phase is influenced by sequential occurrences, with each event relying on tasks.

Author 1: Arpita Singh
Author 2: Sanjay K. Singh
Author 3: Hari Kiran Vege
Author 4: Nilu Singh

Keywords: Cyber-crime; digital forensics; digital evidence; data analysis; security and privacy; cyber-attack

PDF

Paper 40: Deep Learning and Classification Algorithms for COVID-19 Detection

Abstract: The imaging modalities of chest X-rays and computed tomography (CT) are commonly utilized to quickly and accurately diagnose COVID-19. Due to time and human error, it is exceedingly difficult to manually identify the infection using radio imaging. COVID-19 identification is being mechanized and improved with the use of artificial intelligence (AI) tools that have already showed promise. This study employs the following methodology: The chest footage was pre-processed by setting equalizing the histogram, sharpening it, and so on. The transformed chest images are then retrieved through shallow and high-level feature mapping over the backbone network. To further improve the classification performance of the convolutional neural network, the model uses self-attained mechanism through feature maps. Numerous simulations show that CT image classification and augmentation may be accomplished with higher efficiency and flexibility using the Inception-Resnet convolutional neural network than with traditional segmentation methods. The experiment illustrates the association between model accuracy, model loss, and epoch. Inception-statistical Resnet's measurement results are 98%, 91%, 91%.

Author 1: Mohammed Sidheeque
Author 2: P. Sumathy
Author 3: Abdul Gafur. M

Keywords: Deep Learning; COVID-19; classification; artificial intelligence

PDF

Paper 41: Gamification on OTT Platforms: A Behavioural Study for User Engagement

Abstract: This study examines the consumer’s visual attention toward gamification options while watching the OTT (Over-the-top) online content. Also, the impact of gamification on user engagement (UE) on the OTT platform was studied using data collected by conducting an eye-tracking experiment and subsequently using a user engagement scale (UES). The study was carried out at the marketing and behavioural lab of a management institute in India using the OTT platform website and Tobii eye-tracker. Empirical data was collected from 52 respondents within the age group between 23 to 35 years. The relation between Attention to Gamification (AG), Reward Satisfaction (RS), and User Engagement (UE) were studied by running a mediating linear regression analysis. From the results, it was found that respondents were equally interested in watching the online content as well as ready to explore the gamification options. The research findings demonstrate that Reward Satisfaction (RS) acted as a mediating factor in the relation between Attention to Gamification (AG) and User Engagement (UE). This study adds to the literature on consumer engagement towards gamification on the OTT platform, where the literature is still limited. Future research could consider mobile apps as a platform to undertake the study. This study aimed to empirically test the effect of AG on UE with the involvement of RS as a mediator. The study is the first of its type to use eye-tracking data to understand the impact of gamification on the OTT platform.

Author 1: Komal Suryavanshi
Author 2: Prasun Gahlot
Author 3: Surya Bahadur Thapa
Author 4: Aradhana Gandhi
Author 5: Ramakrishnan Raman

Keywords: Gamification; user engagement; eye-tracking; OTT (Over-the-top); reward; visual attention

PDF

Paper 42: Optimally Allocating Ambulances in Delhi using Mutation based Shuffled Frog Leaping Algorithm

Abstract: This paper presents a reliable and competent evolutionary-based approach for improving the response time of Emergency Medical Service (EMS) by efficiently allocating ambulances at the base stations. As the prime objective of EMS is to save people's lives by providing them with timely assistance, thus increasing the chances of a person's survivability, this paper has undertaken the problem of ambulance allocation. The work has been implemented using the proposed mutation-based Shuffled Frog Leaping Algorithm (mSFLA) to provide an optimal allocation plan. The authors have altered the basic SFLA using the concept of mutation to improve the quality of the solution obtained and avoid being trapped in local optima. Considering a set of assumptions, the new algorithm has been applied for allocating 50 ambulances among 11 base stations in Southern Delhi. The working environment of EMS, which includes stochastic requests, travel time, and dynamic traffic conditions, has been considered to attain accurate results. The work has been implemented in the MATLAB simulation environment to find an optimized allocation plan with a minimum average response time. The authors have reduced the average response time by 12.23% with the proposed algorithm. The paper also compares mSFLA, Genetic Algorithm (GA), and Particle Swarm Optimization (PSO) for the stated problem. The algorithms are compared in terms of objective value (average response time), convergence rate, and constancy repeatability to conclude that mSFLA performs better than the other two algorithms.

Author 1: Zaheeruddin
Author 2: Hina Gupta

Keywords: Ambulance allocation; ambulance service; emergency medical service; shuffled frog leaping algorithm; mutation based shuffled frog leaping algorithm

PDF

Paper 43: Adopting a Digital Transformation in Moroccan Research Structure using a Knowledge Management System: Case of a Research Laboratory

Abstract: Digital Transformation has become one of the most discussed debates; many sectors have adopted digital transformation to gain a competitive advantage and to ensure their continuity. Moroccan universities, in their turn, are facing strategic and managerial challenges due to emerging practices related to digital transformation. To address this issue, the proposed work sets out to define the factors that lead us to adopt a digital transformation using SWOT analysis and to apply total quality management techniques to contribute to our research laboratory's digital transformation, by digitalizing and managing knowledge and processes. KMS-TQM digital platform has been used to capitalize knowledge and profile the different existing functions, positions, tasks, and referential competencies. Then, we analyzed all the actual processes to propose a business process re-engineering using Bizagi Modeler. The study’s contribution is to standardize all the current processes in the laboratory to help the Doctoral Studies Center successfully carry out the digital transformation. Moreover, the aim is to make all functions and tasks for each position explicit.

Author 1: Fatima-Ezzahra AIT-BENNACER
Author 2: Abdessadek AAROUD
Author 3: Khalid AKODADI
Author 4: Bouchaib CHERRADI

Keywords: Business process re-engineering; digital transformation; knowledge management system; Moroccan research laboratory; total quality management

PDF

Paper 44: Analyzing the Relationship between the Personality Traits and Drug Consumption (Month-based user Definition) using Rough Sets Theory

Abstract: There is no doubt that the use of drugs has significant consequences for society, it introduces risk into the human life and causing earlier mortality and morbidity. Being a conscientious member of society, we must go ahead to prevent these young minds from life-threatening addiction. Owing to the computational complexity of wrapper approaches, the poor performance of filtering techniques, and the classifier dependency of embedded approaches, artificial intelligence and machine learning systems can provide useful tools for raising the prediction rate of drug users. Recently, the psychologists approved the recent personality traits Five Factor Model (FFM) for understanding human individual differences. The aim of this work is to propose a rough sets theory based method to investigate the relationship between drug user/non-user (month-based user definition) and the personality traits. The data of five factor personality profiles, impulsivity, sensation-seeking and biographical information of users of 21 different types of legal and illegal drugs are used to fetch all reducts and finally a set of classification rules are created to predict the drug user/non-user(month-based user definition). The outcomes demonstrate the novelty of the current work which can be summarized as The set of generalized classification rules which pronounced with logic functions build a knowledge base with excellent accuracy to analyze drug misuse successfully and may be worthy in many applications.

Author 1: Manasik M. Nour
Author 2: H. A. Mohamed
Author 3: Sumayyah I. Alshber

Keywords: Classification; personality traits; five factor model; rules extraction; drug abuse detection; rough sets theory; feature selection

PDF

Paper 45: Wavelet Multi Resolution Analysis based Data Hiding with Scanned Secrete Images

Abstract: Wavelet Multi Resolution Analysis (MRA) based data hiding with scanned secrete images is proposed for improvement of invisibility of the secrete images. Daubechies (biorthogonal basis was adopted as the wavelet, but it was demonstrated that the key image (or secret image data) information can be restored with the biorthogonal wavelet. Also, the information of what to adopt as the biorthogonal wavelet is hidden. Key image information can also be protected by doing so, that the horizontal biorthogonal wavelet of the image does not have to be the same as the vertical biorthogonal wavelet, and the insertion position of the secret image data can be freely selected. It is also possible to divide the bit string of the secret image data and insert it into an arbitrary high frequency component, that the information hiding capability changes depending on the number of bit strings (information amount) of the secret image data, and the secret image in the public image data. Random scanning is effective for improving the visibility of data, selection of scanning method type, random number initial value It was shown that sharing only among parties is useful for improving confidentiality, resistance to noise, resistance to data compression, and resistance to tampering with data.

Author 1: Kohei Arai

Keywords: Multi-Dimensional wavelet transformation; multi resolution analysis: MRA; image data hiding; scanned secrete image; Daubechies basis function; invisibility

PDF

Paper 46: Multiple Eye Disease Detection using Hybrid Adaptive Mutation Swarm Optimization and RNN

Abstract: The major cause of visual impairment in aged people is due to age related eye diseases such as cataract, diabetic retinopathy, and glaucoma. Early detection of eye diseases is necessary for better diagnosis. This paper concentrates on the early identification of various eye disorders such as cataract, diabetic retinopathy, and glaucoma from retinal fundus images. The proposed method focuses on the automated early detection of multiple diseases using hybrid adaptive mutation swarm optimization and regression neural networks (AED-HSR). In the proposed work, the input images are preprocessed and then multiple features such as entropy, mean, color, intensity, standard deviation, and statistics are extracted from the collected data. The extracted features are segmented by using an adaptive mutation swarm optimization (AMSO) algorithm to segment the disease sector from the fundus image. Finally, the features collected are fed to a regression neural network (RNN) classifier to classify each fundus image as normal or abnormal. If the classifier output is abnormal, then it is classified by the corresponding diseases in terms of cataract, glaucoma, and diabetic retinopathy, which improves the accuracy of detection and classification. Ultimately, the results of the classifiers are evaluated by several performance analyses and the viability of structural and functional features is considered. The proposed system predicts the type of the disease with an accuracy of 0.9808, specificity of 0.9934, sensitivity of 0.9803 and F1 score of 0.9861 respectively.

Author 1: P. Glaret Subin
Author 2: P. Muthu Kannan

Keywords: Adaptive mutation swarm optimization; fundus image; feature extraction; RNN classifier; standard deviation

PDF

Paper 47: Insect Pest Image Detection and Classification using Deep Learning

Abstract: Farmers' primary concern is to reduce crop loss because of pests and diseases, which occur irrespective of the cultivation process used. Worldwide more than 40% of the agricultural output is lost due to plant pathogens, insects, and weed pests. Earlier farmers relied on agricultural experts to detect pests. Recently Deep learning methods have been utilized for insect pest detection to increase agricultural productivity. This paper presents two deep learning models¬¬ based on Faster R-CNN Efficient Net B4 and Faster R-CNN Efficient Net B7 for accurate insect pest detection and classification. We validated our approach for 5, 10, and 15 class insect pests of the IP102 dataset. The findings illustrate that our proposed Faster R-CNN Efficient Net B7 model achieved an average classification accuracy of 99.00 %, 96.00 %, and 93.00 % for 5, 10, and 15 class insect pests outperforming other existing models. To detect insect pests less computation time is required for our proposed Faster-R-CNN method. The investigation reveals that our proposed Faster R-CNN model can be used to identify crop pests resulting in higher agricultural yield and crop protection.

Author 1: Niranjan C Kundur
Author 2: P B Mallikarjuna

Keywords: Deep learning; faster RCNN; insect pest detection; IP102 dataset; efficient net

PDF

Paper 48: Analysis of Noise Removal Techniques on Retinal Optical Coherence Tomography Images

Abstract: In the biomedical field, automatic disease detection by image processing has become the norm in the current days. For early illness detection, ophthalmologists have explored a variety of invasive and noninvasive procedures. Optical Coherence Tomography (OCT) is a noninvasive imaging technique for obtaining high resolution tomographic images of biological systems. The image quality is degraded by noise, which degrades the performance of noisy image processing algorithms. The OCT images captured with speckle noise and prior to further processing, it is critical to use an effective approach for denoising the image. In this paper, we used Median filter, Average filter or Mean filter, Wiener filter, Gaussian filter and Bilateral filter on OCT images in this paper, and discussed the advantages and drawbacks of each approach. The effectiveness of these filters are compared using the Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE) and Contrast to Noise Ratio (CNR).

Author 1: T. M. Sheeba
Author 2: S. Albert Antony Raj

Keywords: Average or Mean filter; Bilateral filter; denoising image; Gaussian Filter; Median filter; optical coherence tomography; Wiener filter

PDF

Paper 49: Analyzing the State of Mind of Self-quarantined People during COVID-19 Pandemic Lockdown Period: A Multiple Correspondence Analysis Approach

Abstract: COVID-19 (Corona) virus has spread across the world threatening lives of millions of people. In India first COVID-19 case was detected on 30th January 2020 in Kerala. To minimize the spread of Corona Virus, countries all over the world implemented complete lockdown. Due to complete lockdown even people who are not exposed to corona virus, have to self-quarantine to keep themselves safe from getting infected by the disease. People (especially Indians) have never experienced such complete lockdown and quarantining situations before. Thus, this creates a space for curiosity that how people are going to react to this situation. The present study aims to analyse how self-quarantined people during COVID-19 lockdown period are reacting to quarantining, what measures they are taking to deal with this situation, and what are their sentiments towards quarantining. The study also aims to measure their Happiness and to identify the factors that are statistically significant to Happiness. For this study, the data is collected through a survey method. Multiple correspondence analysis are performed to analyse the survey data. The happiness score is evaluated by using the GNH (Gross National Happiness) methodology. Proportional Odd Logistics Regression is used to identify factors that are statistically significant in predicting happiness. The study suggests that family factor is related to the happiness of the self-quarantined people during such lockdown situations.

Author 1: Gauri Vaidya
Author 2: Vidya Kumbhar
Author 3: Sachin Naik
Author 4: Vijayatai Hukare

Keywords: Correspondence analysis; happiness index; sentiment analysis; proportional odds logistic regression; self-quarantining

PDF

Paper 50: SIBI (Sign System Indonesian Language) Text-to-3D Animation Translation Mobile Application

Abstract: This research proposed a mobile application prototype to translate Indonesian text into SIBI (Sign System for the Indonesian Language) 3D gestures animation to bridge the communication gap between the deaf and the other. To communicate in sign language, the signer will use his/her hands and fingers to demonstrate the word gesture, and at the same time, his/her mouth will pronounce the word being expressed. Therefore, the proposed mobile application needs two animation generator components: the hand gesture and the lip movement generator. Hand gestures are made using a motion capture sensor. Mouth movements are created for all syllables available in the SIBI dictionary using the Dirichlet Free-Form Deformation (DFFD) method. The subsequent challenging work is synchronizing these two components and adding transitional gestures. A transitional gesture done by the cross-fading method is needed to make a word gesture that can smoothly connect with the next word gesture. The Mean Opinion Score (MOS) test was run to measure the mouth movements in 3D animation. The MOS score is 4.422. There are four surveys conducted to measure user satisfaction. The surveys showed that the animation generated did not significantly differ from the original video. The Sistem Usability Score (SUS) is 76.25. The score means that prototype is in the GOOD category. The average time needed to generate an animation from Indonesian input text is less than 100ms.

Author 1: Erdefi Rakun
Author 2: Sultan Muzahidin
Author 3: IGM Surya A. Darmana
Author 4: Wikan Setiaji

Keywords: SIBI sign language; sequence generation; visual speech; animation

PDF

Paper 51: A Review of Foreground Segmentation based on Convolutional Neural Networks

Abstract: Foreground segmentation in dynamic videos is a challenging task for many researchers. Many researchers worked on various methods that were traditionally developed; however, the performance of those state-of-art procedures has not yielded encouraging results. Hence, to obtain efficient results, a deep learning-based neural network model is proposed in this paper. The proposed methodology is based on Convolutional Neural Network (CNN) model incorporated with Visual Geometry Group (VGG) 16 architecture, which is further divided into two sections, namely, Convolutional Neural Network section for feature extraction and Transposed Convolutional Neural Network (TCNN) section for un-sampling feature maps. Then the thresholding technique is employed for effective segmentation of foreground from background in images. The Change Detection (CDNET) 2014 benchmark dataset is used for the experimentation. It consists of 11 categories, and each category contains four to six videos. The baseline, camera jitter, dynamic background, and bad weather are the categories considered for the experimentation. The performance of the proposed model is compared with the state-of-the-art techniques, such as Gaussian Mixture Model (GMM) and Visual Background Extractor (VIBE) for its efficiency in segmenting foreground images.

Author 1: Pavan Kumar Tadiparthi
Author 2: Sagarika Bugatha
Author 3: Pradeep Kumar Bheemavarapu

Keywords: Foreground segmentation; deep learning; Convolutional Neural Network (CNN); Visual Geometry Group (VGG) 16 architecture; Transposed Convolutional Neural Network (TCNN); Gaussian Mixture Model (GMM); Visual Background Extractor (VIBE)

PDF

Paper 52: Multi-instance Finger Knuckle Print Recognition based on Fusion of Local Features

Abstract: Personal identity has become an important asset in today's digital world for any individual in society. Biometrics offers itself as a reliable and secure guarantor of our identities, so it has become essential to build efficient and robust recognition systems. In this orientation, we propose a fusion approach, which aims to optimally exploit the dividing block dimensions in the case of local methods to reduce similarities. We will use the compound local binary model (CLBP) for local features extraction, a robust operator descriptor that exploits both the sign and the inclination information of the differences between the center and the neighbor gray values. The reliability of the proposed approach was evaluated on the PolyU Finger Knuckle Print (FKP) database. We presented several experimental results that show the detailed path of our approach, explain the choices made for each step and illustrate the significant improvements compared to other existing recognition systems in the literature. The recognition rate of the proposed global approach is one of the highest among the other methods. Optimal final approach recognition rates vary between 99.70% and 100%.

Author 1: Amine AMRAOUI
Author 2: Mounir AIT KERROUM
Author 3: Youssef FAKHRI

Keywords: Biometrics; Finger Knuckle Print; local features; fusion; compound local binary pattern

PDF

Paper 53: Detection of Credit Card Fraud using a Hybrid Ensemble Model

Abstract: The rising number of credit card frauds presents a significant challenge for the banking industry. Many businesses and financial institutions suffer huge losses because card users are reluctant to use their cards. A primary goal of fraud detection is to identify prior transaction patterns to detect future fraud. In this paper, a hybrid ensemble model is proposed to combine bagging and boosting techniques to distinguish between fraudulent and legitimate transactions. During the experimentation two datasets are used; the European credit card dataset and the credit card stimulation dataset which are highly imbalanced. The oversampling method is used to balance both datasets. To overcome the problem of unbalanced data oversampling method is used. The model is trained to predict output results by combining random forest with Adaboost. The proposed model provides 98.27 % area under curve score on the European credit cards dataset and the stimulation credit card dataset gives 99.3 % area under curve score.

Author 1: Sayali Saraf
Author 2: Anupama Phakatkar

Keywords: Credit card; hybrid ensemble model; bagging; boosting; data imbalance

PDF

Paper 54: Covid-19 and Pneumonia Infection Detection from Chest X-Ray Images using U-Net, EfficientNetB1, XGBoost and Recursive Feature Elimination

Abstract: The pandemic caused by the COVID-19 virus is the most serious current threat to the public's health. For the purpose of identifying patients with Covid-19, Chest X-Rays have proven to be an indispensable imaging modality for the hospital. Nevertheless, radiologists are needed to commit a significant amount of time to their interpretation. It is possible to diagnose and triage cases of Covid-19 effectively and rapidly with the assistance of precise computer systems that are powered by Machine Learning techniques. Machine Learning techniques such as Deep Feature Extraction can help detect the disease with improved precision and speed when used in conjunction with X-Ray images of the lung. This helps to alleviate the problem of lack of testing kits. Using the U-Net for Semantic image segmentation for lung segmentation and deep feature extraction-based strategy that was suggested in this research, it is possible to differentiate between patients who have contracted the Covid-19 virus, pneumonia and healthy people. XGBoost and recursive feature extraction based proposed methodology is evaluated using 20 different Pre-Trained deep learning based models including EfficientNet variations and it is observed that the maximum detection accuracy, precision, recall specificity, and F1-score are achieved when EfficientNetB1 is used to extract deep features. The respective values for these metrics are 97.6%, 0.964, 0.964, and 0.982. These findings lend credence to the efficiency of the proposed methodology.

Author 1: Munindra Lunagaria
Author 2: Vijay Katkar
Author 3: Krunal Vaghela

Keywords: Covid-19; u-net; efficientnet; semantic image segmentation; XGboost; recursive feature extraction

PDF

Paper 55: Analysis of Privacy and Security Challenges in e-Health Clouds

Abstract: Electronic Health Records (EHR) techniques are being used at an increasingly faster rate to store the patient data making it easier to retrieve, share and utilize it efficiently. This data can be used for research purposes, clinical trials and for studying epidemiology to come up with strategies for epidemic control. With a huge global inflation, the increasing costs of healthcare and the shortage of medicine, it becomes convenient for the healthcare organizations to migrate from the traditional healthcare system to a more sophisticated, cost effective and efficient cloud-based e-Health model. To optimize the full potential of an ICT-based e-Health system, it is imperative for the existing healthcare systems to be implemented in a full-fledged cloud environment. However, with numerous benefits of technology, it might pose some privacy and security threats as well. Therefore, the security and access control of such information is of vital significance. Nonetheless, with the increasing interest of healthcare organizations to migrate from the conventional healthcare systems to the modern cloud-based e-Health systems, not much care is being taken to address security and privacy issues effectively towards the protection of sensitive data.

Author 1: Reem Alanazi

Keywords: HER; e-health; security; privacy; cloud

PDF

Paper 56: Identification of Retinal Disease using Anchor-Free Modified Faster Region

Abstract: Infections of the retinal tissue, as well as delayed or untreated therapy, may result in visual loss. Furthermore, when a large dataset is involved, the diagnosis is prone to inaccuracies. As a consequence, a completely automated model of retinal illness diagnosis is presented to eliminate human input while maintaining high accuracy classification findings. ODALAs (Optimal Deep Assimilation Learning Algorithms) are unable to handle zero errors or covariance or linearity and normalcy. DLTs (Deep Learning Techniques) such as GANs (Generative Adversarial Networks) or CNNs might replace the numerical solution of dynamic systems (Convolution Neural Networks), in order to speed up the runs. With this objective, this study proposes a completely automated multi-class retina disorders prediction system in which pictures from the Fundus image dataset are upgraded using RSWHEs (Recursive Separated Weighted Histogram Equalizations) to boost contrast and noise is eliminated using the Wiener filter. The improved picture is used for segmentation, which is done using clustering and the optimum threshold. The suggested EFFCM is used for clustering (Enriched Fast Fuzzy C Means). The suggested AOO (Adaptive optimum Otsu) threshold technique is used for clustering and picture optimal thresholding. This work suggests AMF-RCNNs (anchor-free modified faster region-based CNNs) that integrate AFRPNs (anchor free regions proposal generation networks) with Improved Fast R-CNNs into single networks for detecting retinal issues accurately. The performances of Accuracy is 98.5%, F-Measure is 96.5%, Precession is 99.2% and different Subset features are 98.5 % show better results when compared with other related techniques or models.

Author 1: Arulselvam. T
Author 2: S. J. Sathish Aaron Joseph

Keywords: Retinal disease; fundus image dataset; contrast enhancement; segmentation; Fast Fuzzy C Means; adaptive optimal OTSU; faster region-based convolutional neural network

PDF

Paper 57: OpenCV Implementation of Grid-based Vertical Safe Landing for UAV using YOLOv5

Abstract: The challenge of proving autonomous landing in practical situations is difficult and highly risky. Adopting autonomous landing algorithms substantially minimizes the probability of human-involved mishaps, which may enable the use of drones in populated metropolitan areas to their full potential. This paper proposes an Unmanned Aerial Vehicles (UAV) vertical safe landing & navigation pipeline that relies on lightweight computer vision modules, able to execute on the limited computational resources on-board a typical UAV. In this work, a grid-based mask technique is proposed for selecting the safe landing zones where each grid is parameterizable based on the size of the UAVs, which is implemented using OpenCV. A custom trained YOLOv5 model is the underlying building block for safe landing algorithm which is trained for aerial views of pedestrians, cars & bikes to identify as obstacles. The nearest obstacle-free zone algorithm is applied over the YOLOv5 output where boundary box locations are identified using Hue Saturation Value (HSV) filtering and then split into grids for safe landing zones where maximum coverage is taken into account while analyzing each scene. It performs a 2-level operation to prevent collisions while descending at different altitudes. Since UAV is expected to be processing only at predetermined altitudes, which will shorten the processing time, generating a PID signal for UAV actuators to navigate to the required safe zone with utmost safety and accuracy.

Author 1: Hrusna Chakri Shadakshri V
Author 2: Veena M. B
Author 3: Keshihaa Rudra Gana Dev V

Keywords: Autonomous UAV system; computer vision algorithm; YOLOv5; safe landing site selection; Haversine equations

PDF

Paper 58: Gaussian Projection Deep Extreme Clustering and Chebyshev Reflective Correlation based Outlier Detection

Abstract: Outlier detection or simply the task of point detection that are noticeably distinct and different from data sample is a predominant issue in deep learning. When a framework is constructed, these distinctive points can later lead to model training and compromise accurate predictions. Owing to this reason, it is paramount to recognize and eliminate them before constructing any supervised model and this is frequently the initial step when dealing with a deep learning issue. Over the recent few years, different numbers of outlier detector algorithms have been designed that ensure satisfactory results. However, their main disadvantages remain in the time and space complexity and unsupervised nature. In this work, a clustering-based outlier detection called, Random Projection Deep Extreme Learning-based Chebyshev Reflective Correlation (RPDEL-CRC) is proposed. First, Gaussian Random Projection-based Deep Extreme Learning-based Clustering model is designed. Here, by applying Gaussian Random Projection function to the Deep Extreme Learning obtains the relevant and robust clusters corresponding to the data points in a significant manner. Next, with the robust clusters, outlier detection time is said to be reduced to a greater extent. In addition, a novel Chebyshev Temporal and Reflective Correlation-based Outlier Detection model is proposed to detect outliers therefore achieving high outlier detection accuracy. The proposed approach is validated with the NIFTY-50 stock market dataset. The performance of the RPDEL-CRC method is evaluated by applying it to NIFTY-50 Stock Market dataset. Finally, we compare the results of the RPDEL-CRC method to the state-of-the-art outlier detection methods using outlier detection time, accuracy, error rate and false positive rate evaluation metrics.

Author 1: S. Rajalakshmi
Author 2: P. Madhubala

Keywords: Outlier detection; clustering; Gaussian random projection; deep extreme learning; Chebyshev distance; temporal; reflective correlation

PDF

Paper 59: Efficient Decentralized Sharing Economy Model based on Blockchain Technology: A Case Study of Najm for Insurance Services Company

Abstract: Blockchain is an emerging technology that is used to address ownership, centrality, and security issues in different fields. The blockchain technology has converted centralized applications into decentralized and distributed ones. In existing sharing economy applications, there are issues related to low efficiency and high complexity of services. However, blockchain technology can be adopted to overcome these issues by effectively opening up secure information channels of the sharing economy industry and other related parties, encouraging industry integration and improving the ability of sharing economy organizations to readily gain required information. This paper discusses blockchain technology to enhance the development of insurance services by proposing a five-layer decentralized model using Ethereum platform. The Najm for Insurance Services Company in Saudi Arabia was employed in a case study for applying the proposed model to effectively solve the issue of online underwriting, and to securely and efficiently enhance the verification and validation of transactions. The paper concludes with a review of the lessons learned and provides suggestions for blockchain application development process.

Author 1: Atheer Alkhammash
Author 2: Kawther Saeedi
Author 3: Fatmah Baothman
Author 4: Rania Anwar Aboalela
Author 5: Amal Babour

Keywords: Blockchain; decentralized; Ethereum; multichain; Najm; sharing economy

PDF

Paper 60: Virtual Communities of Practice to Promote Digital Agriculturists’ Learning Competencies and Learning Engagement: Conceptual Framework

Abstract: Virtual Communities of Practice (VCoPs) are networks of people who share a common interest and a desire to learn together in the same domain via ICT. The limitations of the existing concepts for developing VCoPs in general contexts are not explained in terms of the integration between virtual learning technologies and digital learning strategies used to promote expected learning outcomes in the agricultural sector. This research aims to propose the conceptual framework for developing the Virtual Communities of Practice through Digital Inquiry (VCoPs-DI Model) to promote digital agriculturists’ learning competencies and engagement. The research methodology was divided into three stages: the first stage involves a literature review for document analysis and synthesis, the second stage involves constructing the conceptual framework, and the third stage involves evaluating the content validity index. The key results showed that the developed conceptual framework has three parts: (1) The fundamentals of concept formation were divided into four concept bases: (1.1) Communities of Practice (CoPs), (1.2) Virtual Learning Environments (VLEs), (1.3) Digital Learning Resources (DLRs), and (1.4) Critical Inquiry Method; (2) The identification of the manipulated variable was divided into two compositions: (2.1) VCoPs and (2.2) Digital Inquiry (DI); (3) The identification of the dependent variable was divided into two compositions: (3.1) Digital agriculturists' learning competencies, and (3.2) learning engagement. Findings from an expert’s review show that the scale levels of the content validity index (SCVI) were 0.958. We anticipate that our conceptual framework could be used for reference as part of the design and development of the VCoPs model to promote learning in the agricultural sector.

Author 1: Maneerat Manyuen
Author 2: Surapon Boonlue
Author 3: Jariya Neanchaleay
Author 4: Vitsanu Nittayathammakul

Keywords: VCoPs; digital inquiry; digital agriculture; learning competencies; learning engagement

PDF

Paper 61: Deep Q-learning Approach based on CNN and XGBoost for Traffic Signal Control

Abstract: Traffic signal control is a way for reducing traffic jams in urban areas, and to optimize the flow of vehicles by minimizing the total waiting times. Several intelligent methods have been proposed to control the traffic signal. However, these methods use a less efficient road features vector, which can lead to suboptimal controls. The objective of this paper is to propose a deep reinforcement learning approach as the hybrid model that combines the convolutional neural network with eXtreme Gradient Boosting to traffic light optimization. We first introduce the deep convolutional neural network architecture for the best features extraction from all available traffic data and then integrated the extracted features into the eXtreme Gradient Boosting model to improve the prediction accuracy. In our approach; cross-validation grid search was used for the hyper-parameters tuning process during the training of the eXtreme Gradient Boosting model, which will attempt to optimize the traffic signal control. Our system is coupled to a microscopic agent-based simulator (Simulation of Urban MObility). Simulation results show that the proposed approach improves significantly the average waiting time when compared to other well-known traffic signal control algorithms.

Author 1: Nada Faqir
Author 2: Chakir Loqman
Author 3: Jaouad Boumhidi

Keywords: Convolutional neural network; extreme gradient; traffic control; traffic optimization; urban mobility

PDF

Paper 62: Automatic Text Summarization using Document Clustering Named Entity Recognition

Abstract: Due to the rapid development of internet technology, social media and popular research article databases have generated many open text information. This large amount of textual information leads to 'Big Data'. Textual information can be recorded repeatedly about an event or topic on different websites. Text summarization (TS) is an emerging research field that helps to produce summary from a single or multiple documents. The redundant information in the documents is difficult, hence part or all of the sentences may be omitted without changing the gist of the document. TS can be organized as an exposition to collect accents from its special position, rather than being semantic in nature. Non-ASCII characters and pronunciation, including tokenizing and lemmatization are involved in generating a summary. This research work has proposed an Entity Aware Text Summarization using Document Clustering (EASDC) technique to extract summary from multi-documents. Named Entity Recognition (NER) has a vital part in the proposed work. The topics and key terms are identified using the NER technique. Extracted entities are ranked with Zipf’s law and sentence clusters are formed using k-means clustering. Cosine similarity-based technique is used to eliminate the similar sentences from multi-documents and produce unique summary. The proposed EASDC technique is evaluated using CNN dataset and it shown an improvement of 1.6 percentage when compared with the baseline methods of Textrank and Lexrank.

Author 1: Senthamizh Selvan. R
Author 2: K. Arutchelvan

Keywords: Named entity recognition; text summarization; k-means clustering; Zipf’s law

PDF

Paper 63: Convolutional Neural Networks with Transfer Learning for Pneumonia Detection

Abstract: Pneumonia is a type of acute respiratory infection caused by microbes, and viruses that affect the lungs. Pneumonia is the leading cause of infant mortality in the world, accounting for 81% of deaths in children under five years of age. There are approximately 1.2 million cases of pneumonia in children under five years of age and 180 000 died in 2016. Early detection of pneumonia can help reduce mortality rates. Therefore, this paper presents four convolutional neural network (CNN) models to detect pneumonia from chest X-ray images. CNNs were trained to classify X-ray images into two types: normal and pneumonia, using several convolutional layers. The four models used in this work are pre-trained: VGG16, VGG19, ResNet50, and InceptionV3. The measures that were used for the evaluation of the results are Accuracy, recall, and F1-Score. The models were trained and validated with the dataset. The results showed that the Inceptionv3 model achieved the best performance with 72.9% accuracy, recall 93.7%, and F1-Score 82%. This indicates that CNN models are suitable for detecting pneumonia with high accuracy.

Author 1: Orlando Iparraguirre-Villanueva
Author 2: Victor Guevara-Ponce
Author 3: Ofelia Roque Paredes
Author 4: Fernando Sierra-Liñan
Author 5: Joselyn Zapata-Paulini
Author 6: Michael Cabanillas-Carbonell

Keywords: Neural networks; transfer learning; pneumonia; detection; Convolutional

PDF

Paper 64: A Monadic Co-simulation Model for Cyber-physical Production Systems

Abstract: The production flexibility required by the new industrial revolutions is largely based on heterogeneous Cyber-Physical Production Systems models that cooperate with each other to perform complex tasks. To accomplish tasks at an acceptable pace, CPPSs should be based on appropriate cooperation mechanisms. To this end a CPPS must be able to provide services in the form of functionalities to other CPPSs, and also to use functionalities of other CPPSs. The cooperation of two CPPS systems is done by co-simulating the two models that allow the partial or total access of the functionalities of one system, by the other system. Requests from one CPPS to another CPPS create connection moments of the two models that can only be performed in certain states of the two models. Also, the answers to these requests create connections between the two models in other subsequent states. Optimal aggregation of the behaviors of the two models, by co-simulation, is essential because otherwise it can lead to very long waiting times and can cause major problems if not done correctly. We will see in this paper that the behavior of such a simulation model can be represented by a category, and the co-simulation of two models can be defined by a monad determined by two adjoint functors between the simulation categories of the two models.

Author 1: Daniel-Cristian Craciunean

Keywords: Models; metamodels; co-simulation; adjoint functors; monads; cyber-physical production systems

PDF

Paper 65: A Machine Learning Model for Predicting Heart Disease using Ensemble Methods

Abstract: There is the continuous increase in death rate related to cardiac disease across the world. Prediction of the heart disease in advance may help the experts to suggest the pre-emptive measures to minimize the death risk. The early diagnosis of heart disease symptoms is made possible by machine learning technologies. The existing machine learning models are inefficient in terms of simulation error, accuracy and timing for heart disease prediction. Hence, an efficient approach is needed for efficient prediction of heart disease. In the current research paper, a model based on Machine learning techniques has been proposed for early and accurate prediction of heart disease. The proposed model is based on techniques for feature optimization, feature selection, and ensemble learning. Using WEKA 3.8.3 tool, the feature selection and feature optimisation technique has been applied for irrelevant features elimination and then the pragmatic features are tested using ensemble techniques. Further, the comparison of the proposed model is made with the existing model without feature selection and feature optimisation technique in terms of heart disease prediction effectiveness. It is found that the results of proposed model gives the better performance in terms of simulation error, response time and accuracy in heart disease prediction.

Author 1: Jasjit Singh Samagh
Author 2: Dilbag Singh

Keywords: Heart disease; diagnosis; ensemble; optimization; prediction

PDF

Paper 66: Novel Approach in Classification and Prediction of COVID-19 from Radiograph Images using CNN

Abstract: Effective screening and early detection of COVID-19 patients are highly crucial to slow down and stop the disease's rapid spread at this time. Currently, RT-PCR, CT scanning and Chest X-ray (CXR) imaging are the diagnosis mechanisms for COVID-19 detection. In this proposed work radiology examination by using CXR images is used for COVID-19 detection due to dearth of CT Scanners and RT-PCR testing centers. Therefore, researchers have developed various Deep and Machine Learning systems that can predict COVID-19 using CXR images. Out of which, few are exhibited good prediction results. However, Most of the models are suffered with over fitting, high variance, memory and generalization errors which are caused by noise as well as datasets are limited. Therefore, a Convolutional Neural Network (CNN) with the leveraging Efficient Net architecture is proposed for COVID-19 case detection. The proposed methods have an accuracy of 99% which gives the better results than the present available methods. Therefore, the proposed model can be used in real-time covid-19 classification systems.

Author 1: Chalapathiraju Kanumuri
Author 2: CH. Renu Madhavi
Author 3: Torthi Ravichandra

Keywords: COVID-19; x-ray images; deep learning technique; CNN; efficient net

PDF

Paper 67: Machine Learning based Electromigration-aware Scheduler for Multi-core Processors

Abstract: The rising performance demands in modern technology devices see the need to pack more functionality per area and are made possible with the advent of technology scaling. The extremely down-scaled, high-density processors used in such technology devices functioning at high frequencies and greater temperatures expedite various aging effects which impact the reliable lifetime of computing systems. Electromigration is considered to be an important intrinsic aging effect that reduces the useful lifetime of modern microprocessors. The objective of this work is to use machine learning methods to develop an electromigration-aware scheduler for assigning workloads to cores based on reliability and performance requirements. Aging estimation of the processor cores is performed based on the proposed computationally efficient and accurate regression-based thermal prediction models. According to experimental findings, the suggested technique can significantly extend the lifetime of multi-core architectures while allowing performance to degrade gracefully. The maximum error in the prediction of the lifetime of the cores using the proposed methodology is estimated to be 2.85%.

Author 1: Jagadeesh Kumar P
Author 2: Mini M G

Keywords: Electromigration aware scheduler; useful lifetime; multi-core processor reliability; machine learning model

PDF

Paper 68: Sentiment Analysis on Acceptance of New Normal in COVID-19 Pandemic using Naïve Bayes Algorithm

Abstract: The COVID-19 pandemic has such a significant impact and causes difficulties in many aspects that the new normal rules should be implemented to reduce the effects. New normal rules have been implemented by governments worldwide to break the virus chain and stop its transmission among the society. Even if the COVID-19 outbreak is under control, governments still need to know whether society could adapt and adjust to their new daily lifestyles. Many precautions still must be addressed as the transition to endemic status does not mean that COVID-19 will naturally eventually disappear. The World Health Organization also has warned that it is too early to treat COVID-19 as an endemic disease. Since the pandemic, many interactions have been done online, leading to the increasing social media usage to express opinions about COVID-19. The objective of the study is to explore the capability of the Naïve Bayes algorithm in the sentiment classification of the public’s acceptance on the new normal in the COVID-19 pandemic. Naïve Bayes has been chosen for its good performance in solving various other classification problems. In this study, Twitter data were used for the analysis and were collected between March and June 2022. The evaluation results have shown that Naïve Bayes could generate excellent and acceptable performance in the classification with an accuracy of 83%. According to the findings of this research, many people have accepted the new normal in their daily lives. The future works would include scrapping more data based on geolocation, improving the feature extraction technique, balancing the dataset and comparing Naïve Bayes performance with other well-known classifiers. The subsequent study could also focus on detecting the emotions of the public and processing non-English tweets.

Author 1: Siti Hajar Aishah Samsudin
Author 2: Norlina Mohd Sabri
Author 3: Norulhidayah Isa
Author 4: Ummu Fatihah Mohd Bahrin

Keywords: Sentiment analysis; COVID-19; new normal; acceptance; naïve bayes

PDF

Paper 69: Partial Differential Equation (PDE) based Hybrid Diffusion Filters for Enhancing Noise Performance of Point of Care Ultrasound (POCUS) Images

Abstract: A hybrid filter is developed by combining smoothing and edge preservation properties of anisotropic diffusion (AD) filters and noise reduction features of median filtering. Mixed Gaussian Impulse noise and speckle noise are considered for analysis. The performance of this hybrid filter is verified using ultrasound images. The effectiveness of this filter is assessed with Point of Care Ultrasound (POCUS) images to verify whether the algorithm developed is applicable to them. POCUS refers to a handheld portable ultrasound instrument that can be used at patient bedside. Quantitative analysis with COVID-19 POCUS images, in terms of SNR, SSIM and MSE is performed. Results demonstrate that for all test images, the proposed filter has the best SNR, least MSE, and highest SSIM. Significant improvement in image quality is thus observed both qualitatively and quantitatively. The novelty of suggested technique is its effectiveness in reducing both mixed Gaussian impulse noise and speckle noise in ultrasound as well as POCUS images without the need for separate filters. POCUS has played a significant role in the diagnosis and management of pulmonary, cardiac and vascular pathologies associated with COVID-19. Automatic segmentation of these images and subsequent automatic detection and diagnosis are becoming increasingly popular due to the rapid development of artificial intelligence technologies. These results are useful in implementing better pre-processing prior to segmentation of ultrasound images to facilitate improved patient care.

Author 1: Deepa V S
Author 2: Jagathyraj V P
Author 3: Gopikakumari R

Keywords: Anisotropic diffusion filter; POCUS; mixed Gaussian impulse noise; speckle noise

PDF

Paper 70: Smart Greenhouse Monitoring and Controlling based on NodeMCU

Abstract: Food security is one of the major rising issues as the human population is larger and the land available for cultivation is smaller, as well unassured affairs happened often in society especially in the current CoVID-19 rapidly spread days. To mitigate this condition, further improve the yields and quality of food, this paper proposed a smart and low-cost greenhouse monitoring and control system, which mainly consists of sensors, actuators, LCD display and microcontrollers. DHT22 sensor is used to get the surrounding temperature and humidity in the greenhouse, and NodeMCU is used as the main microcontroller. Some other facilities such as fan and heater are used to adjust the inside environment. The system could monitor the growth environment continuously with Internet-connected, the monitoring data is transmitted and stored in the ThingSpeak cloud, the users can visualize the live data through a webpage or phone APP in real-time. If the environment condition is out of the predefined level, the environment is monitored continuously, and the system can be adjusted automatically. This system can be deployed in the greenhouse simply and maintain the greenhouse environment in a normal range dynamically and continuously.

Author 1: Yajie Liu

Keywords: Smart greenhouse; ThingSpeak cloud; NodeMCU

PDF

Paper 71: Design of Accounting Information System in Data Processing: Case Study in Indonesia Company

Abstract: This study aims to determine the implementation of System Application and Product in Data Processing (SAP) in a company to provide solutions for companies to obtain reliable reports and improve performance in a company. This study uses the mixed method through interviews with resource persons who have work in well-known company. The data obtained were analyzed by the method of literature study from data on the internet. The results of this study indicate that many companies still apply manual systems in reporting, one of them is the lack of adequate technological facilities within the company so that companies cannot fulfill their business processes optimally due to not using integrated system that connected with each other.

Author 1: Meiryani
Author 2: Dezie Leonarda Warganegara
Author 3: Agustinus Winoto
Author 4: Gabrielle Beatrice Hudayat
Author 5: Erna Bernadetta Sitanggang
Author 6: Ka Tiong
Author 7: Jessica Paulina Sidauruk
Author 8: Mochammad Fahlevi
Author 9: Gredion Prajena

Keywords: Accounting; information systems; SAP; ERP; implementation of SAP

PDF

Paper 72: MOOC Dropout Prediction using FIAR-ANN Model based on Learner Behavioral Features

Abstract: Massive Open Online Courses (MOOCs) are a transformative technology in digital learning that incorporates new techniques through video sessions, exams, activities, and conversations. Everyone leads a successful life in their professional and personal skills learning courses during COVID-19. The research concentrated on employing video interaction analysis to characterize crucial MOOC tasks, including predicting dropouts and student achievement. Our work consists of merely generating and picking the best characteristics based on the learner behavior for evaluating the dropout measure. To locate the frequent objects for feature creation, an association rule-FP growth approach is applied. The neural network is implemented using frequent itemset-3, which is used for feature selection. The evaluation metrics are calculated by using the Multilayer Perceptron (MLP) method. The metric values were then compared to the proposed model and some base supervised machine learning models namely Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), K-Nearest Neighbor (KNN), and Naive Bayes (NB). The FIAR (Feature Importance Association Rule)-ANN(Artificial Neural Network) dropout prediction model was tested on the KDD Cup 2015 dataset and it had a high accuracy of over 92.42, which is approximately 18% better than the MLP-NN model. With the optimized parameters, we are solely focused on lowering dropout rates and increasing learner retention.

Author 1: S. Nithya
Author 2: S.Umarani

Keywords: Dropout prediction; data analytics; association rule mining; machine learning; artificial neural network

PDF

Paper 73: Sentiment Analysis of Online Movie Reviews using Machine Learning

Abstract: Many websites encourage their users to write reviews for a wide variety of products and services. In particular, movie reviews may influence the decisions of potential viewers. However, users face the arduous tasks of summarizing the information in multiple reviews and determining the useful and relevant reviews among a very large number of reviews. Therefore, we developed machine learning (ML) models to classify whether an online movie review has positive or negative sentiment. We utilized the Stanford Large Movie Review Dataset to build models using decision trees, random forests, and support vector machines (SVMs). Further, we compiled a new dataset comprising reviews from IMDb posted in 2019 and 2020 to assess whether sentiment changed owing to the coronavirus disease 2019 (COVID-19) pandemic. Our results show that the random forests and SVM models provide the best classification accuracies of 85.27% and 86.18%, respectively. Further, we find that movie reviews became more negative in 2020. However, statistical tests show that this change in sentiment cannot be discerned from our model predictions.

Author 1: Isaiah Steinke
Author 2: Justin Wier
Author 3: Lindsay Simon
Author 4: Raed Seetan

Keywords: Decision tree; machine learning (ML); natural language processing (NLP); random forests; sentiment analysis; support vector machine (SVM); reviews

PDF

Paper 74: Detection and Extraction of Faces and Text Lower Third Techniques for an Audiovisual Archive System using Machine Learning

Abstract: As part of the audiovisual archive digitization project, which has become a complex field that requires human and material resources, and its automation and optimization have so far represented a center of interest for researchers and media manufacturers, in particular those linked to the integration of artificial intelligence tools in the industry, an elaborate work for the development of an optical character and face recognition model, to digitize the tasks of audiovisual archivist from the manuscript method in automation, from a TV news video. In this article, an approach to develop an example of lower third in Arabic language and facial detection and recognition for news presenter that provide accurate classification results as well as the presentation of different methods and algorithms for Arabic characters. Many studies have been presented in this area, however a satisfactory classification accuracy is yet to be achieved. The comparative state-of-the-art results adopt the latest approaches to study face recognition or OCR, but this model supports both at the same time. it will present the context of realization, the method proposed to extract the texts in the video, using machine learning, about the specificity of the Arabic language, and finally the reasons that govern the decisions taken in the steps of realization. The best results from this approach in real project at the media station was 90.60%. The dataset collected via presenters images and the character dataset via the Pytesseract library.

Author 1: Khalid El Fayq
Author 2: Said Tkatek
Author 3: Lahcen Idouglid
Author 4: Jaafar Abouchabaka

Keywords: Image processing; OpenCV; Tesseract; video OCR; face detection

PDF

Paper 75: Data Recovery Comparative Analysis using Open-based Forensic Tools Source on Linux

Abstract: Data recovery is one of the forensic techniques used to recover data that has been lost or deleted. Data recovery is carried out if there is a condition where the data that has been owned is deleted or damaged. If the data has been lost or deleted or even tampered with, then a forensic expert has several ways to restore data that has been lost or damaged. One of them is to use a complete data recovery method using forensic tools, namely, TSK Recover, FTK Imager, Foremost Recover, and Testdisk Recover. Unfortunately, tools such as FTK imager and TSK recover have a weakness, namely that some damaged or corrupted data files cannot be restored in their entirety; they can only be recovered but not be opened. This study uses a tool comparison method approach using foremost recover and Testdisk recover. It's just that this method cannot be used using the graphic user interface (GUI) but using the CLI (Command Line) in the LINUX operating system. And the files that have been recovered will be fully recovered.

Author 1: Muhammad Fahmi Abdillah
Author 2: Yudi Prayudi

Keywords: Recovery; tools; FTK imager; foremost; Testdisk

PDF

Paper 76: Advanced Persistent Threat Attack Detection using Clustering Algorithms

Abstract: Advanced Persistent Threat (APT) attack has become one of the most complex attacks. It targets sensitive information. Many cybersecurity systems have been developed to detect the APT attack from network data traffic and request. However, they still need to be improved to identify this attack effectively due to its complexity and slow move. It gets access to the organizations either from an active directory or by gaining remote access, or even by targeting the Domain Name Server (DNS). Nowadays, many machine learning (ML) techniques have been implemented to detect APT attack by using the tools in the market. However, still, there are some limitations in terms of accuracy, efficiency, and effectiveness, especially the lack of labeled data to train ML methods. This paper proposes a framework to detect APT attacks using the most applicable clustering algorithms, such as the APRIORI, K-means, and Hunt’s algorithm. To evaluate and compare the performance of the proposed framework, several experiments are conducted on a public dataset. The experimental results showed that the Support Vector Machine with Radial Basis Function (SVM-RBF) achieves the highest accuracy rate, reaching about 99.2%. This accurate result confirms the effectiveness of the developed framework for detecting attacks from network data traffic.

Author 1: Ahmed Alsanad
Author 2: Sara Altuwaijri

Keywords: APT Attack detection; DNS; network; cybersecurity; clustering algorithms

PDF

Paper 77: Energy Efficient Node Deployment Technique for Heterogeneous Wireless Sensor Network based Object Detection

Abstract: Lifetime of the network and the quality of operation are the two important issues in a wireless sensor network system meant for object detection and tracking application. At the same time, there should be a tradeoff between network cost and the quality of operation as high cost of the network limits its real-time usability. Heterogeneous wireless sensor networks promises the prolonged network lifetime as well as enhances network reliability as they contain a mixture of nodes with different characteristics. Further prolongation of network lifetime can be achieved by managing the available node energy in a proper way, i.e, by minimizing number of communication, minimizing node density, minimizing overhead information generated during operation etc. Proper node deployment scheme not only helps to enhance the lifetime of the network but also helps in reducing deployment cost while maintaining the quality of operation in terms of object detection accuracy. This paper focuses on the energy efficient node deployment in heterogeneous wireless sensor network system with the features of maximum network coverage, optimum node density and optimum network cost. This paper proposes a novel energy efficient node deployment algorithm that determines the number of static and mobile nodes required for deployment and then relocates the mobile nodes to cover up the coverage hole using 8-neighbourhood and Particle Swarm Optimization (PSO) algorithm. The performance of the proposed algorithm is compared with corresponding model of Harmony Search Algorithm (HSA) and PSO based node deployment and it is seen that the proposed model outperforms better in comparison to them.

Author 1: Jayashree Dev
Author 2: Jibitesh Mishra

Keywords: Heterogeneous wireless sensor network; energy efficiency; node deployment; object detection network; particle swarm optimization; harmony search algorithm

PDF

Paper 78: Fine-grained Access Control in Distributed Cloud Environment using Trust Valuation Model

Abstract: Cloud computing has been in existence as an adaptable technology that gets integrated with IoT, Big-Data, and WSN to provide reliable, scalable and mesh-free services. However, because of its openness in nature, the privacy of the cloud is an important parameter for today’s research. The most important privacy factor in cloud is access control and user trust. Many articles related to access control and trust management were presented, but most of them include highly complex algorithms that result in network overhead. This proposed security framework is for a better and more effective system wherein multiple distributed centers are created with trust-based computing for authentication and validation of requests from users and their resources. The idea of trust here is for efficient decision-making and establishing reliable relationships among users and resources using least computations. Each user has different permissions for each file present in the cloud server. The simulated results shows improvement in the rate of successful transactions, time cost and network overhead.

Author 1: Aparna Manikonda
Author 2: Nalini N

Keywords: Fine-grained; distributed; access control; trust

PDF

Paper 79: BERT-Based Hybrid RNN Model for Multi-class Text Classification to Study the Effect of Pre-trained Word Embeddings

Abstract: Due to the Covid-19 pandemic which started in the year 2020, many nations had imposed lockdown to curb the spread of this virus. People have been sharing their experiences and perspectives on social media on the lockdown situation. This has given rise to increased number of tweets or posts on social media. Multi-class text classification, a method of classifying a text into one of the pre-defined categories, is one of the effective ways to analyze such data that is implemented in this paper. A Covid-19 dataset is used in this work consisting of fifteen pre-defined categories. This paper presents a multi-layered hybrid model, LSTM followed by GRU, to integrate the benefits of both the techniques. The advantages of word embeddings techniques like GloVe and BERT have been implemented and found that, for three epochs, the transfer learning based pre-trained BERT-hybrid model performs one percent better than GloVe-hybrid model but the state-of-the-art, fine-tuned BERT-base model outperforms the BERT-hybrid model by three percent, in terms of validation loss. It is expected that, over a larger number of epochs, the hybrid model might outperform the fine-tuned model.

Author 1: Shreyashree S
Author 2: Pramod Sunagar
Author 3: S Rajarajeswari
Author 4: Anita Kanavalli

Keywords: Multi-class text classification; transfer learning; pre-training; word embeddings; GloVe; bidirectional encoder representations from transformers; long short-term memory; gated recurrent units; hybrid model; RNN

PDF

Paper 80: A Hybrid Approach of Wavelet Transform, Convolutional Neural Networks and Gated Recurrent Units for Stock Liquidity Forecasting

Abstract: Stock liquidity forecasting is critical for investors, issuers, and financial market regulators. The object of this study is to propose a method capable of accurately predicting the liquidity of stocks. The few studies on stock liquidity forecasting have focused on single models such as Seasonal Auto-Regressive Integrated Moving Average with eXogenous factors, the nonlinear autoregressive network with exogenous input, and Deep Learning. A new trend in forecasting which attempts to combine several approaches is emerging at the moment. Inspired by this new trend, we propose a hybrid approach of Wavelet Transform, Convolutional Neural Networks, and Gated Recurrent Units to predict stock liquidity. Our model is tested on daily data of companies listed on the Casablanca Stock Exchange from 2000 to 2021. Its forecasting performances are evaluated based on the Mean Absolute Error, the Root Mean Square Error, the Mean Absolute Percentage Error, Theil’s U statistic, and the correlation coefficient. Finally, the outperformance of the proposed model is confirmed by comparison with other reference forecasting models. This study contributes to the enrichment of the field of prediction of financial risks and can constitute a framework of analysis allowing to help the stakeholders of the financial markets to forecast the liquidity of the actions.

Author 1: Mohamed Ben Houad
Author 2: Mohammed Mestari
Author 3: Khalid Bentaleb
Author 4: Adnane El Mansouri
Author 5: Salma El Aidouni

Keywords: Stock liquidity; wavelet transform; convolutional neural networks; GRU cell; Casablanca stock exchange

PDF

Paper 81: Visual Navigation System for Autonomous Drone using Fiducial Marker Detection

Abstract: Drones have been quickly developing for civilian applications in recent years. Because of the nonlinearity of the mathematical drone model, and the importance of precise navigation to avoid possible dangers, it is necessary to establish an algorithm to localize the drone simultaneously and maneuver it to the desired destination. This paper presents a visual-based multi-stage error tolerance navigation algorithm of an autonomous drone by a tag-based fiducial marker detection in finding its target. Dynamic and kinematic models of the drone were developed by Newton-Euler. The position and orientation of the drone, related to the tag, are determined by AprilTag, which is used as feedback in a closed-loop control system with an Adjustable Proportional-Integral-Derivative (APID) controller. Parameters of the controller are tuned based on steady-State error, which is defined as the distance of the drone from the desired point. The sequence of path trajectory, that drone follows to reach the desired point, is defined as a navigation algorithm. A model of the drone was simulated in a virtual outdoor to mimic hovering in complex obstacles environment. The results present satisfactory performance of the navigation system programmed by the APID controller in comparison with the conventional Proportional-Integral-Derivative (PID) controller. It can be ascertained that the proposed navigation system based on a tag marker in the closed-loop control system is applicable to maneuvering the drone autonomously and useful for various industrial tasks in indoor/outdoor environments.

Author 1: Mohammad Soleimani Amiri
Author 2: Rizauddin Ramli

Keywords: Proportional-Integral-Derivative (PID) controller; AprilTag detection system; autonomous navigation; fiducial marker detection

PDF

Paper 82: Design of a Mobile Application for the Logistics Process of a Fire Company

Abstract: Currently, the logistics process is an important part for any company because it helps to manage the assets and products that enter and leave it. Some companies carry out this process physically, saving the information on sheets of paper or Excel files, which takes longer to do and is not at the forefront of how companies do it, which is by using mobile applications to improve this process. Likewise, it has been decided to implement a mobile application with the aim of improving the logistics process in the Callao No. 15 fire company. For the elaboration of the application, the RUP methodology was used to do it in a more optimal way, in the end, a survey of experts in Google Forms was conducted, addressed to 10 experts to know the evaluation of the mobile application. In the end, a favorable result was obtained from the opinion of the experts on the mobile application; 70% of the respondents indicate that the usability of the mobile application has a “Very high” level;it can be seen that 80% of respondents indicate that the presentation of the mobile the application has a “Very high” level; it can be seen that 90%of the respondents indicate that the functionality of the mobile application has a “Very high” level; besides,it can be seen that 80% of the respondents indicate that the security of the mobile application has a “Very high” level.

Author 1: Luis Enrique Parra Aquije
Author 2: Luis Gustavo Vasquez Carranza
Author 3: Gustavo Bernnet Alfaro Pena
Author 4: Michael Cabanillas-Carbonell
Author 5: Laberiano Andrade-Arenas

Keywords: Fire company; logistics process; mobile application; RUP methodology; expert survey

PDF

Paper 83: Intelligent System for Personalised Interventions and Early Drop-out Prediction in MOOCs

Abstract: In this paper, we propose an approach to early detect students at high risk of drop-out in MOOC (Massive Open Online Course); we design personalised interventions to mitigate that risk. We apply Machine Learning (ML) algorithms and data mining techniques to a dataset extracted from XuetangX MOOC learning platforms and sourced from the KDD cup 2015. Since this dataset contains only raw student log activity records, we perform a hybrid feature selection and dimensionality reduction techniques to extract relevant features, and reduce models complexity and computation time. Besides, we built two models based on: Genetic Algorithms (GA) and Deep Learning (DL) with supervised learning methods. The obtained results, according to the accuracy and the AUC (Area Under Curve)-ROC (Reciever Operator Characteristic) metrics, prove the pertinence of the extracted features and encourage the use of the hybrid features selection. They also proved that GA and DL are outperforming the baseline algorithms used in related works. To assess the generalisation of the approach used in this work, The same process is performed to a second benchmark dataset extracted from the university MOOC. Then, a single web application hosted on the university server, produces an individual weekly drop-out probability, using time series data. It also proposes an approach to personalise and prioritise interventions for at-risk students according to the drop-out patterns.

Author 1: ALJ Zakaria
Author 2: BOUAYAD Anas
Author 3: Cherkaoui Malki Mohammed Oucamah

Keywords: MOOC; drop-out; dimensionality reduction; features selection; personalised intervention

PDF

Paper 84: An Intelligent Decision Support Ensemble Voting Model for Coronary Artery Disease Prediction in Smart Healthcare Monitoring Environments

Abstract: Coronary Artery Disease (CAD) is one of the most common cardiac diseases worldwide and causes disability and economic burden. It is the world’s leading and most serious cause of mortality, with approximately 80% of deaths reported in low- and middle-income countries. The preferred and most precise diagnostic tool for CAD is angiography, but it is invasive, expensive, and technically demanding. However, the research community is increasingly interested in the computer-aided diagnosis of CAD via the utilization of machine learning (ML) methods. The purpose of this work is to present an e-diagnosis tool based on ML algorithms that can be used in a smart healthcare monitoring system. We applied the most accurate machine learning methods that have shown superior results in the literature to different medical datasets such as RandomForest, XGboost, MultilayerPerceptron, J48, AdaBoost, NaiveBayes, LogitBoost, KNN. Every single classifier can be efficient on a different dataset. Thus, an ensemble model using majority voting was designed to take advantage of the well-performed single classifiers, Ensemble learning aims to combine the forecasts of multiple individual classifiers to achieve higher performance than individual classifiers in terms of precision, specificity, sensitivity, and accuracy; furthermore, we have bench-marked our proposed model with the most efficient and well-known ensemble models, such as Bagging, Stacking methods based on the cross-validation technique, The experimental results confirm that the ensemble majority voting approach based on the top three classifiers: MultilayerPerceptron, RandomForest, and AdaBoost, achieves the highest accuracy of 88,12% and outperforms all other classifiers. This study demonstrates that the majority voting ensemble approach proposed above is the most accurate machine learning classification approach for the prediction and detection of coronary artery disease.

Author 1: Anas Maach
Author 2: Jamila Elalami
Author 3: Noureddine Elalami
Author 4: El Houssine El Mazoudi

Keywords: Machine learning; smart healthcare; coronary artery disease

PDF

Paper 85: Estimation of Varying Reaction Times with RNN and Application to Human-like Autonomous Car-following Modeling

Abstract: The interaction between human-driven vehicles and autonomous vehicles has become a vital issue in micro-transportation science. Compared to autonomous vehicles, human-driven vehicles have varying reaction times that could compromise traffic efficiency and stability. But human drivers can anticipate future traffic conditions subconsciously, which guar-antees qualified performance. This paper proposes an estimation method of varying reaction times and a human-like autonomous car-following model. The varying reaction times are estimated based on recurrent neural networks (RNNs) after the cross-correlation analysis of human-driven vehicles’ trajectory profiles. A human-like autonomous car-following model is established based on Intelligent Driver Model (IDM), considering both varying reaction times and temporal anticipation, and the short form is IDM RTTA. The analytical string stability of IDM RTTA is deduced and illustrated. The trajectory simulation result shows that increasing accuracy of trajectory prediction is obtained with the proposed model, which will benefit the interaction between human-driven vehicles and autonomous vehicles.

Author 1: Lijing Ma
Author 2: Shiru Qu
Author 3: Junxi Zhang
Author 4: Xiangzhou Zhang

Keywords: Car-following model; intelligent driver model; human-driven vehicle; autonomous vehicle; varying reaction time; string stability

PDF

Paper 86: Mobile Food Journalling Application with Convolutional Neural Network and Transfer Learning: A Case for Diabetes Management in Malaysia

Abstract: Diabetes is an ever worsening problem in modern society, placing a heavy burden on healthcare systems. Due to the association between obesity and diabetes, food journaling mobile applications are an effective approach for managing and improving the outcome of diabetics. Due to the efficacy of nutritional tracking and management in managing diabetes, we implemented a deep learning-based Convolutional Neural Network food classification model to aid with food logging. The model is trained on a subset of the Food-101 and Malaysian Food 11 datasets, including web-scraped images, with a focus on food items found locally in Malaysia. In our experiments, we explore how fine-tuning of the image dataset improves the performance of the model.

Author 1: Jason Thomas Chew
Author 2: Yakub Sebastian
Author 3: Valliapan Raman
Author 4: Patrick Hang Hui Then

Keywords: Convolutional neural network; deep learning; diabetes; food journal; mobile application; nutritional tracking; Malaysia

PDF

Paper 87: Rethinking Classification of Oriented Object Detection in Aerial Images

Abstract: With the help of the rapid development of technology, especially the prevalence of UAVs (unmanned aerial vehicles), object detection in aerial images gains much more attention in computer vision and deep learning. However, traditional methods use horizontal bounding boxes for object representation leading to inconsistency between objects and features. Therefore, many detectors are being built to tackle this problem, and normally they use the conventional approaches of training and testing to achieve the results. Our pipeline proposed to strengthen not only the classification but also localization via independent training processes using convex-hull transformation in data pre-processing phase. We experimented with the well-designed S2ANet, R3Det, ReDet, RoI Transformer and Oriented R-CNN on the well-established oriented object detection dataset DOTA. Then we adopt the best detectors with the well-known classification network EfficientNet to our proposed pipeline and achieve promising results on the oriented object detection DOTA dataset. Moreover, our pipeline can flexibly be adapted to various oriented object detection baselines improving the results in classification via independent extensive training cycles.

Author 1: Phuc Nguyen
Author 2: Thang Truong
Author 3: Nguyen D. Vo
Author 4: Khang Nguyen

Keywords: Aerial images; classification; convex-hull transformation; data processing; oriented object detection

PDF

Paper 88: TextBrew: Automated Model Selection and Hyperparameter Optimization for Text Classification

Abstract: In building a machine learning solution, algorithm selection and hyperparameter tuning is the most time-consuming task. Automated Machine Learning is a solution to fully automate the process of finding the best model for a given task without actually having to try various models. This paper introduces a new AutoML system, TextBrew, explicitly built for the NLP task of text classification. Our system provides an automated method for selecting transformer models, tuning hyperparameters, and combining the best models into one by ensembling. Keeping in mind that new state-of-the-art models are being constantly introduced, TextBrew has been designed to be highly flexible and thus can support additional models easily. In our work, we experiment with multiple transformer models, each with numerous different hyperparameter settings, and select the most robust models. These models are then trained on multiple datasets to obtain accuracy scores, which are then used to build the meta-dataset to train the meta-model. Since text classification datasets are not as abundant, our system generates synthetic data to augment the meta-dataset using CopulaGAN, a deep generative model. The meta-model is an ensemble of five models, which predicts the best candidate model with an accuracy of 78.75%. The final model returned to the user is an ensemble of all the best models that can be trained under the given time constraint. Experiments on various datasets and comparisons with existing systems demonstrate the effectiveness of our system.

Author 1: Rushil Desai
Author 2: Aditya Shah
Author 3: Shourya Kothari
Author 4: Aishwarya Surve
Author 5: Narendra Shekokar

Keywords: Automated machine learning; AutoML; NLP; trans-former models; hyperparameter optimization; CopulaGAN; gener-ative model; meta-learning

PDF

Paper 89: On-Device Major Indian Language Identification Classifiers to Run on Low Resource Devices

Abstract: Language Identification acts a first and necessary step in building intelligent Natural Language Processing (NLP) systems that handle code mixed data. There is a lot of work around this problem, but there is still scope for improvement, especially for local Indian languages. Also, earlier works mostly concentrates on just accuracy of the model and neglects the information like, whether they can be used on low resource devices like mobiles and wearable devices like smart watches with considerable latency. Here, this paper discusses about both binary classification and multiclass classification using character grams as the features. Considering total nine languages in this classification which includes, eight code mixed Indian languages with English (Hindi, Bengali, Kannada, Tamil, Telugu, Gujarati, Marathi, Malayalam) and standard English. Binary classifier discussed in this paper will classify Hinglish (Hindi when written using English script is commonly known as Hinglish) from seven other code-mixed Indian Languages with English and standard English. Multiclass classifier will classify the previously mentioned languages. Binary classifier gave an accuracy of 96% on the test data and the size of the model was 1.4 MB and achieved an accuracy of 87% with multiclass classifier on same test set with model size of 3.6 MB.

Author 1: Yashwanth Y S

Keywords: Character grams; code-mixed; deep learning; Indian languages; language identification; NLP; social media text

PDF

Paper 90: Evaluating Hybrid Framework of VASNET and IoT in Disaster Management System

Abstract: During emergency operations in Disaster Management System (DMS) for natural and man-made disasters, any breakdown in the existing information and communication technology will affect the aspect of effectiveness and efficiency on an emergency response task. For Vehicular Ad-hoc Sensor Network (VASNET), the limitation in terms of infrastructure that consists of RSU (Roadside Sensor Unit) may partially or fully destroy the post-disaster scenario. As such, performance degradation of VASNET affects the network infrastructure on high packet loss, delay, and produce a huge amount of energy consumption in DMS. Thus, modification of VASNET and integrate with Internet of Thing (IoT) technology is a must to improve and solving the current problem on VASNET technology. Therefore, the main objective of this study was to investigate the performance of the proposed modified VASNET framework integrated with IoT at DMS in terms of energy consumption and packet loss. A suggested node in the proposed framework was introduced to implement low data rate and high data rate in evaluating the proposed framework using LTE and LTE-A transmission protocol. It was found that LTE-A contributes more energy by 25.33 (mJ/Byte) compared to LTE on 20 (mJ/Byte) on a high data rate. On the other hand, in terms of low data rate, LTE-A influences the most on the proposed framework by recording 19.82(mJ/Byte), LTE only 19.33 (mJ/Byte). For packet loss, LTE shows a high packet loss rate by contributing 11.39% compared to LTE-A, which is 8.0% in terms of low data rate, and 14.80% compared to LTE-A, only 11.97% for high data rate. Consequently, LTE-A on high data rate contributes more energy consumption and LTE in packet loss on same data rate.

Author 1: Sia Chiu Shoon
Author 2: Mohammad Nazim Jambli
Author 3: Sinarwati Mohamad Suhaili
Author 4: Nur Haryani Zakaria

Keywords: Energy consumption; packet loss; LTE-A; VASNET; IoT

PDF

Paper 91: A Novel Machine Learning-based Framework for Detecting Religious Arabic Hatred Speech in Social Networks

Abstract: Social media platforms generate a huge amount of data every day. However, liberty of speech through these networks could easily help in spreading hatred. Hate speech is a severe concern endangering the cohesion and structure of civil societies. With the increase in hate and sarcasm among the people who contact others over the internet in this era, there is a dire need for utilizing artificial intelligence (AI) technology innovation that would face this problem. The rampant spread of hate can dangerously break society and severely damage marginalized people or groups. Thus, the identification of hate speech is essential and becoming more challenging, where the recognition of hate speech on time is crucial in stopping its dissemination. The capacity of the Arabic morphology and the scarcity of resources for the Arabic language makes the task of distinguishing hate speech even more demanding. For fast identification of Arabic hate speech in social network comments, this work presents a comprehensive framework with eight machine learning (ML) and deep learning (DL) algorithms, namely Gradient Boosting (GB), K-Nearest Neighbor (K-NN), Logistic Regression (LR), Naive Bayes (NB), Passive Aggressive Classifier (PAC), Support Vector Machine (SVM), Ara-BERT, and BERT-AJGT are implemented. Two representation techniques have been used in the proposed framework in order to extract features: a bag of words followed by BERT-based context text representations. Based on the result and discussion part, context text representation techniques with Ara-BERT and BERT-AJGT outperform all other ML models and related work with accuracy equal to 79% for both models.

Author 1: Mahmoud Masadeh
Author 2: Hanumanthappa Jayappa Davanager
Author 3: Abdullah Y. Muaad

Keywords: Machine learning; Arabic language; hatred detection; social network; classification algorithm

PDF

Paper 92: Modeling Multioutput Response Uses Ridge Regression and MLP Neural Network with Tuning Hyperparameter through Cross Validation

Abstract: The multiple regression model is very popular among researchers in both field of social and science because it is easy to interpret and have a well-established theoretical framework. However, the multioutput multiple regression model is actually widely applied in the engineering field because in the industrial world there are many systems with multiple outputs. The ridge regression model and the Multi-Layer Perceptron (MLP) neural network model are representations of the predictive linear regression model and predictive non-linear regression model that are widely applied in the world of practice. This study aims to build multi-output models of a ridge regression model and an MLP neural network whose hyperparameters are determined by a grid search algorithm through the cross-validation method. The hyperparameter that produces the smallest RMSE value in the validation data is chosen as the hyperparameter to train both models on the training data. The hyperparameter in question is a combination of learning algorithms and alpha values (ridge regression), a combination of the number of hidden nodes and gamma values (MLP neural network). In the ridge regression model for alpha in the range between 0.1 and 0.7, the smallest RMSE is obtained for all learning algorithms used. While the MLP neural network model specifically obtained a combination of the number of nodes = 18 and gamma = 0.1 which produces the smallest RMSE. The ridge regression model with selected hyperparameters has better performance (in the RMSE and R2 value) than the MLP neural network model with selected hyperparameters, both on training and testing data.

Author 1: Waego Hadi Nugroho
Author 2: Samingun Handoyo
Author 3: Hsing-Chuan Hsieh
Author 4: Yusnita Julyarni Akri
Author 5: Zuraidah
Author 6: Donna DwinitaAdelia

Keywords: Filter approach; hyperparameter tuning; multi-response; neural network; ridge regression

PDF

Paper 93: Decentralized Access Control using Blockchain Technology for Application in Smart Farming

Abstract: The application of the Internet of Things (IoT) plays a crucial role in the fourth industrial revolution. The sophistication of technology due to the integration of heterogenous smart devices open a new threat from various aspects. Access control is the first line of defence to ensure that IoT resources are secure by preventing illegitimate users from gaining access to these resources. However, access control mechanisms face the limitation of technology in large scale IoT deployments since they are based on a centralized architecture. Significant research concerning decentralized access control solutions for securing IoT resources using combined techniques, such as blockchain, have caught much research attention in recent years. Nevertheless, research for decentralized access control for application in smart farming domain remain as a gap. Thus, this study presented a structured literature review on 81 articles related to the field of access control in IoT and blockchain technology to understand the challenges of centralized access control in securing IoT resources. This study serves as a foundation for decentralized access control using blockchain technology and its application to ensure the IoT actuators and sensors security with the aim to be applied in smart farming. This paper was deliberated based on systematic literature review that was searched from four different database platforms between 2018 and 2021. This study mostly addresses the relevant techniques/approaches including blockchain technology, access control model, key management mechanism and the combination of all three methods. The possible impacts, gap, procedures and evaluation of the decentralized access control are highlighted along with major trends and challenges.

Author 1: Normaizeerah Mohd Noor
Author 2: Noor Afiza Mat Razali
Author 3: Nur Atiqah Malizan
Author 4: Khairul Khalil Ishak
Author 5: Muslihah Wook
Author 6: Nor Asiakin Hasbullah

Keywords: Blockchain; access control; smart contract; internet of thing

PDF

Paper 94: Research on Intelligent Control System of Air Conditioning Based on Internet of Things

Abstract: The current air conditioning intelligent control system cannot achieve the ideal energy-saving effect. The indoor temperature and humidity control is not good enough either. Therefore, an intelligent air-conditioning control system based on Internet of Things technology is designed. The hardware part of the system includes system control motherboard, sensor module, execution control structure, wireless communication module and access layer. The software includes the design of communication layer, the design of monitoring management, and the design of intelligent indoor air-conditioning temperature remote control algorithm. The experimental results show that the control effect of the intelligent air conditioner is more accurate and energy-saving, the opening degree of the air conditioning valve is larger, and the comfort is improved. The indoor temperature and humidity of the proposed system are both more ideal.

Author 1: Binfang Zhang

Keywords: Internet of things technology; intelligent control of air conditioning; system design; double closed-loop load; virtual synchronizer of air conditioning

PDF

Paper 95: A Short Review on the Role of Various Deep Learning Techniques for Segmenting and Classifying Brain Tumours from MRI Images

Abstract: The past few years have observed substantial growth in death rates associated with brain tumors and it is second foremost source of cancer-related demises. However, it is possible to increase the chance of survival if tumors are identified during initial stage by employing various deep learning techniques. These techniques are helpful to the doctors during the diagnosis process. The MRI which refers to magnetic resonance imaging is a non-invasive procedure and low ionization radiation diagnostic tool to evaluate an abnormity that evolves in the form of shape, location or position, size and texture of tumour. This paper focuses on the systematic literature survey of numerous Deep-Learning methods with suitable approaches for tumour segmentation and classification (normal or abnormal) from MRI images. Furthermore, this paper also provides the new aspects of research and clinical solution for brain tumor patients. It incorporates Deep-Learning applications for accurate tumor detection and quantitative investigation of different tumor segmentation techniques.

Author 1: Kumari Kavitha. D
Author 2: E. Kiran Kumar

Keywords: Medical image segmentation; convolutional neural networks (CNN); deep-CNN; feed forward neural networks; brain tumor segmentation (BraTS) and U-net

PDF

Paper 96: Fish Species Classification using Optimized Deep Learning Model

Abstract: Classification of fish species in aquatic pictures is a growing field of research for researchers and image processing experts. Classification of fish species in aquatic images is critical for fish analytical purposes, such as ecological auditing balance, observing fish populations, and saving threatened animals. However, ocean water scattering and absorption of light result in dim and low contrast pictures, making fish classification laborious and challenging. This paper presents an efficient scheme of fish classification, which helps the biologist understand varieties of fish and their surroundings. This proposed system used an improved deep learning-based auto encoder decoder method for fish classification. Optimal feature selection is a major issue with deep learning models generally. To solve this problem efficiently, an enhanced grey wolf optimization technique (EGWO) has been introduced in this study. The accuracy of the classification system for aquatic fish species depends on the essential texture features. Accordingly, in this study, the proposed EGWO has selected the most optimal texture features from the features extracted by the auto encoder. Finally, to prove the efficacy of the proposed method, it is compared to existing deep learning models such as AlexNet, Res Net, VGG Net, and CNN. The proposed method is analysed by varying iterations, batches, and fully connected layers. The analysis of performance criteria such as accuracy, sensitivity, specificity, precision, and F1 score reveals that AED-EGWO gives superior performance.

Author 1: J. M. Jini Mol
Author 2: S. Albin Jose

Keywords: Fish species classification; deep learning; GW optimization; auto encoder decoder; feature selection

PDF

Paper 97: Environmental Noise Pollution Forecasting using Fuzzy-autoregressive Integrated Moving Average Modelling

Abstract: Predicting noise pollution from building sites is important to take precautions to avoid pollution that harms the public. A high accuracy of the prediction model is required so that the predicted model can reach the true value. Forecasting models must be built on solid historical data to achieve high forecasting accuracy. However, data collected through various approaches are subject to ambiguity and uncertainty, resulting in less reliable predictive models. Therefore, the data must be handled accurately, to eliminate data uncertainty. Standard data processing processes are easy to use but do not provide a consistent method for dealing with this ambiguous data. Therefore, a method to deal with data containing uncertainty for forecasting purposes is presented in this paper. A new technique for providing uncertainty-based data preparation has been employed to develop an ARIMA-based model of environmental noise pollution. During the data preparation stage, the standard deviation approach was used. Prior to the development of the prediction model, it is crucial to manage the fuzzy data to minimize errors. The experimental findings show that the suggested data preparation strategy can increase the model's accuracy.

Author 1: Muhammad Shukri Che Lah
Author 2: Nureize Arbaiy
Author 3: Syahir Ajwad Sapuan
Author 4: Pei-Chun Lin

Keywords: Noise pollution; forecasting; ARIMA; uncertainty; standard deviation

PDF

Paper 98: Extractive Multi-document Text Summarization Leveraging Hybrid Semantic Similarity Measures

Abstract: Because of the massive amount of textual information accessible today, automated extraction text summarization is one of the most extensively used ways to organize the information. The summarization mechanisms help to extract the important topics of data from a given set of documents. Extractive summarization is one method for providing a representative summary of a text by choosing the most pertinent sentences from the original text. Extractive multi-document text summarization systems' primary goal is to decrease the quantity of textual information in a document collection by concentrating on the most crucial subjects and removing irrelevant material. In the previous research, there are several methods such as term-weighting schemes and similarity metrics used for constructing an automated summary system. There are few studies that look at the performance of combining various Semantic similarity and word weighting techniques in automatic text summarization. We evaluated numerous semantic similarity metrics in extractive multi-document text summarization in this research. In the extractive multi-document text summarization discussed in this research, we looked at numerous semantic similarity metrics. ROUGE metrics have been used to evaluate the model performance in experiments using DUC datasets. Even more, the combination formed by different semantic similarity measures obtained the highest results in comparison with the other models.

Author 1: Rajesh Bandaru
Author 2: Y. Radhika

Keywords: Extractive text summarization; semantic similarity; sentence scoring; summary

PDF

Paper 99: An Efficient Hybrid LSTM-CNN and CNN-LSTM with GloVe for Text Multi-class Sentiment Classification in Gender Violence

Abstract: Gender-based violence is a public health issue that needs high concern to eliminate discrimination and violence against women and girls. Several cases are through the offline organization and the respective online platform. However, many victims share their experiences and stories on social media platforms. Twitter is one of the methods for locating and identifying gender-based violence based on its type. This paper proposed a hybrid Long Short-Term Memory (LSTM) and Convolution Neural Network CNN with GloVe to perform multi-classification of gender violence. Intimate partner violence, harassment, rape, femicide, sex trafficking, forced marriage, forced abortion, and online violence against women are e eight gender violence keyword for data extraction from Twitter text data. Next is data cleaning to remove unnecessary information. Normalization converts data into a structure the machine can recognize as model input. The evaluation considers cross-entropy loss parameters, learning rate, an optimizer, and epochs. LSTM+GloVe vector embedding outperforms all other methods. CNN-LSTM+Glove and LSTM-CNN+GloVe achieved 0.98 for test accuracy, 0.95 for precision, 0.94 for recall, and 0.95 for the f1-score. The findings can help the public and relevant agencies differentiate and categorize different types of gender violence through text. With this effort, the government can use as one of the mechanisms that indirectly can support monitoring of the current situation of gender violence.

Author 1: Abdul Azim Ismail
Author 2: Marina Yusoff

Keywords: Gender-based violence; deep learning; convolution neural network; long short-term memory; convolution neural network - long short-term memory; long short-term memory - convolution neural network; global vector; multi-class text classification

PDF

Paper 100: Performance Analysis of Deep Learning YOLO models for South Asian Regional Vehicle Recognition

Abstract: For years, humans have pondered the possibility of combining human and machine intelligence. The purpose of this research is to recognize vehicles from media and while there are multiple models associated with this, models that can detect vehicles commonly used in developing countries like Bangladesh, India, etc. are scarce. Our focus was to assimilate the largest dataset of vehicles exclusive to South Asia in addition to the more common universal vehicles and apply it to track and recognize these vehicles, even in motion. To develop this, we increased the class variations and quantity of the data and used multiple variations of the YOLOv5 model. We trained different versions of the model with our dataset to properly measure the degree of accuracy between the models in detecting the more unique vehicles. If vehicle detection and tracking are adopted and implemented in live traffic camera feeds, the information can be used to create smart traffic systems that can regulate congestion and routing by identifying and separating fast and slow-moving vehicles on the road. The comparison between the three different YOLOv5 models led to an analysis that indicates that the large variant of the YOLOv5 architecture outperforms the rest.

Author 1: Minar Mahmud Rafi
Author 2: Siddharth Chakma
Author 3: Asif Mahmud
Author 4: Raj Xavier Rozario
Author 5: Rukon Uddin Munna
Author 6: Md. Abrar Abedin Wohra
Author 7: Rakibul Haque Joy
Author 8: Khan Raqib Mahmud
Author 9: Bijan Paul

Keywords: You Only Look Once (YOLOv5); vehicle detection; neural network; deep learning; vehicle tracking

PDF

Paper 101: Multi-method Approach for User Experience of Selfie-taking Mobile Applications

Abstract: Taking selfies is a popular activity in most social media applications and applying filters/lenses to these selfies has become one of the most demanded features of such applications. This paper aims to design an application for taking selfies to minimize the heavy use of beautifying filters. To understand the current user experience of selfie-taking and filter features, multiple user experience research methods were applied in two steps. In the first step, interviews were conducted with 10 participants to collect data. The key findings of interviews were (i) the need for saving memories as users’ primary goal of using the applications, (ii) the need for using slightly beautifying filters as their preferred filter type, (iii) the need for a favorite filters list, and (iv) the need for the opportunity to edit selfies after they are taken. This output of the interviews was used as an input for determining the survey questions in the second step. A total of 340 respondents completed the survey and the findings were consistent with those of the interviews. Further pointing to the rising opportunity for a new selfie-taking application designed to save selfies quickly without sharing and only apply slightly beautifying filters. More studies should focus on increasing engagement and including a saved selfie categorization feature in the design.

Author 1: Shahad Aldahri
Author 2: Reem Alnanih

Keywords: Filters; lenses; multi-method; research methods; selfie taking; user experience research

PDF

Paper 102: Predicting Academic Performance using a Multiclassification Model: Case Study

Abstract: Now-a-days predicting the academic performance of students is increasingly possible, thanks to the constant use of computer systems that store a large amount of student information. Machine learning uses this information to achieve big goals, such as predicting whether or not a student will pass a course. The main purpose of the work was to make a multiclassifier model that exceeds the results obtained from the machine learning models used independently. For the development of our proposed predictive model, the methodology was used, which consists of several phases. For the first step, 557 records with 25 characteristics related to academic performance were selected, then the preprocessing was applied to said data set, eliminating the attributes with the lowest correlation and those records with inconsistencies, leaving 500 records and 9 attributes. For the transformation, it was necessary to convert categorical to numerical data of four attributes, being the following: SEX, ESTATUS_lab_padre, ESTATUS_lab_madre and CONDITION. Having the data set clean, we proceeded to balance the data, where 1,167 data were generated, using the 2/3 for training and the remaining 1/3 for validation, then the following techniques were applied: Extra Tree, Random Forest, Decision Tree, Ada Boost and XGBoost, each obtained an accuracy of 57.41%, 61.96%, 91.44%, 59.65% and 83.3% respectively. Then the proposed model was applied, combining the five algorithms mentioned above, which reached an accuracy of 92.86%, concluding that the proposed model provides better accuracy than when the models are used independently meaning that it was the one that obtained the best result.

Author 1: Alfredo Daza Vergaray
Author 2: Carlos Guerra
Author 3: Noemi Cervera
Author 4: Erwin Burgos

Keywords: Learning machine; prediction; academic performance; hybrid model; classification techniques; multiclassification; python

PDF

Paper 103: COVID-19 Disease Detection based on X-Ray Image Classification using CNN with GEV Activation Function

Abstract: The globe was rocked by unprecedented levels of disruption, which had devastating effects on daily life, global health, and global economy. Since the COVID-19 epidemic started, methods for delivering accurate diagnoses for multi-category classification have been proposed in this work (COVID vs. normal vs. pneumonia). XceptionNet and Dense Net, two transfer learning pre-trained model networks, are employed in our CNN model. The low-level properties of the two DCNN structures were combined and used to a classifier for the final prediction. To get better results with unbalanced data, we used the GEV activation function (generalized extreme value) to augment the training dataset using data augmentation for validation accuracy, which allowed us to increase the training dataset while still maintaining validation accuracy with the output classifier. The model has been put through its paces in two distinct scenarios. In the first instance, the model was tested using Image Augmentation for train data and the GEV (generalized extreme value) function for output class, and it got a 94% accuracy second instance Model evaluations were conducted without data augmentation and yielded an accuracy rating of 95% for the output class.

Author 1: Karim Ali Mohamed
Author 2: Emad Elsamahy
Author 3: Ahmed Salem

Keywords: COVID-19; CNN; GEV function; image augmentation

PDF

Paper 104: Deep Learning based Cervical Cancer Classification and Segmentation from Pap Smears Images using an EfficientNet

Abstract: One of the most prevalent cancers in the world, cervical cancer claims the lives of many people every year. Since early cancer diagnosis makes it easier for patients to use clinical applications, cancer research is crucial. The Pap smear is a useful tool for early cervical cancer detection, although the human error is always a risk. Additionally, the procedure is laborious and time-consuming. By automatically classifying cervical cancer from Pap smear images, the study's goal was to reduce the risk of misdiagnosis. For picture enhancement in this study, contrast local adaptive histogram equalization (CLAHE) was employed. Then, from this cervical image, features including wavelet, morphological features, and Grey Level Co-occurrence Matrix (GLCM) are extracted. An effective network trains and tests these derived features to distinguish between normal and abnormal cervical images by using EfficientNet. On the aberrant cervical picture, the SegNet method is used to identify and segment the cancer zone. Specificity, accuracy, positive predictive value, Sensitivity, and negative predictive value are all utilized to analyze the suggested cervical cancer detection system performances. When used on the Herlev benchmark Pap smear dataset, results demonstrate that the approach performs better than many of the existing algorithms.

Author 1: Krishna Prasad Battula
Author 2: B. Sai Chandana

Keywords: Cervical cancer; pap smear; time-consuming; contrast local adaptive histogram equalization (CLAHE); Grey Level Co-occurrence Matrix (GLCM); morphological features; wavelet; SegNet

PDF

Paper 105: Cloud based Forecast of Municipal Solid Waste Growth using AutoRegressive Integrated Moving Average Model: A Case Study for Bengaluru

Abstract: Forecasting the quantity of waste growth in upcoming years is very much required for assessing the existing waste management system. In this research work, time series forecast model, ARIMA (Autoregressive Integrated Moving Average), is used to predict future waste growth from 2021 to 2028 for Bengaluru, largest city in Karnataka. Eight years old historical solid waste dataset from 2012 to 2020 is used to make predictions. This dataset is preprocessed and only time bounded variables like days, month, year and waste quantity in tons are used in this research work to obtain accurate prediction. The model is implemented in python in Google Colab free cloud’s Jupyter notebook. As ARIMA is time bounded, forecast made by the model is accurate and performance of the model is evaluated using metrics such as Mean Absolute Deviation (MAD), Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE) and Coefficient of Determination (R2). Outcomes revealed that ARIMA (0, 1, 2) model with the lowermost RMSE (753.5742), MAD (577.4601), and MAPE (11.6484) values and the maximum R2 (0.9788) value has a greater forecast performance. The outcomes attained from the model also showed that the total volume of yearly solid waste to be produced will rise from about 50,300 tons in 2021 to 75,600 tons in 2028.

Author 1: Rashmi G
Author 2: S Sathish Kumar K

Keywords: Cloud Computing; Machine Learning; Time Series Forecasting; Waste Management System; ARIMA; Predictive Modeling

PDF

Paper 106: Building an Intelligent Tutoring System for Learning Polysemous Words in Moore

Abstract: This paper presents the results of our research carried out as part of the building of an Intelligent Tutoring System (ITS) to learn Moor´e, a tone language. A word in tone language may have many meanings according to the pitch. The system has an intelligent tutor to personalize and guide the learning of the transcription of polysemous words in Moor´e. This learning activity aims both to master the transcription and also to distinguish the lexical meaning of words according to the pitch used. A first step of this research has been the specification of the processes, inference and knowledge of the system. In this work we present the implementation and pedagogical assessment of the system. We designed the architecture of the ITS, the diagnosis of transcription errors and remediation approach. Then, we used the Petri net formalism to model the system dynamic in order to analyze its states and fix deadlocks. We developed the system in java and we evaluated its educational value by an experimentation with learners. This shows that the learning objectives can be achieved with this system.

Author 1: Pengwende ZONGO
Author 2: Tounwendyam Frederic OUEDRAOGO

Keywords: Intelligent tutoring system; petri network; evaluation; Moor´e language

PDF

Paper 107: Improving the Diabetes Diagnosis Prediction Rate Using Data Preprocessing, Data Augmentation and Recursive Feature Elimination Method

Abstract: Hyperglycemia is a symptom of diabetes mellitus, a metabolic condition brought on by the body's inability to produce enough insulin and respond to it. Diabetes can damage body organs if it is not adequately managed or detected in a timely manner. Many years of research into diabetes diagnosis has led to a suitable method for diabetes prediction. However, there is still scope for improvement regarding precision. The paper's primary objective is to emphasize the value of data preprocessing, feature selection, and data augmentation in disease prediction. Techniques for data preprocessing, feature selection, and data augmentation can assist classification algorithms function more effectively in the diagnosis and prediction of diabetes. A proposed method is employed for diabetes diagnosis and prediction using the PIMA Indian dataset. A systematic framework for conducting a comparison analysis based on the effectiveness of a three-category categorization model is provided in this study. The first category compares the model's performance with and without data preprocessing. The second category compares the performance of five alternative algorithms employing the Recursive Feature Elimination (RFE) feature selection method. Data augmentation is the third category; data augmentation is done with SMOTE Oversampling, and comparisons are made with and without SMOTE Oversampling. On the PIMA Indian Diabetes dataset, studies showed that data preprocessing, RFE with Random Forest Regression feature selection, and SMOTE Oversampling augmentation can produce accuracy scores of 81.25% with RF, 81.16 with DT, and 82.5% with SVC. From Six Classifiers LR, RF, DT, SVC, GNB and KNN, it is observed that RF, DT, and SVC performed better in accuracy level. The comparative study enables us to comprehend the value of data preprocessing, feature selection, and data augmentation in the disease prediction process as well as how they affect performance.

Author 1: E. Sabitha
Author 2: M. Durgadevi

Keywords: Artificial Intelligence (AI); Machine Learning (ML); Deep Learning(DL); Neural Network; Diabetes Mellitus; Recursive Feature Elimination (RFE); Synthetic Minority Over-sampling Technique (SMOTE)

PDF

Paper 108: A Comparative Study of Unsupervised Anomaly Detection Algorithms used in a Small and Medium-Sized Enterprise

Abstract: Anomaly detection finds application in several industries and domains. The anomaly detection market is growing driven by the increasing development and dynamic adoption of emerging technologies. Depending on the type of supervision, there are three main types of anomaly detection techniques: unsupervised, semi-supervised, and supervised. Given the wide variety of available anomaly detection algorithms, how can one choose which approach is most appropriate for a particular application? The purpose of this evaluation is to compare the performance of five unsupervised anomaly detection algorithms applied to a specific dataset from a small and medium-sized software enterprise, presented in this paper. To reduce the cost and complexity of a system developed to solve the problem of anomaly detection, a solution is to use machine learning (ML) algorithms that are available in one of the open-source libraries, such as the scikit-learn library or the PyOD library. These algorithms can be easily and quickly integrated into a low-cost software application developed to meet the needs of a small and medium-sized enterprise (SME). In our experiments, we considered some unsupervised algorithms available in PyOD library. The obtained results are presented, alongside with the limitations of the research.

Author 1: Irina Petrariu
Author 2: Adrian Moscaliuc
Author 3: Cristina Elena Turcu
Author 4: Ovidiu Gherman

Keywords: Unsupervised anomaly detection algorithms; small and medium-sized enterprise; traceability; open-source libraries

PDF

Paper 109: Automated Brain Disease Classification using Transfer Learning based Deep Learning Models

Abstract: Brain MRI (Magnetic Resonance Imaging) classification is one of the most significant areas of medical imaging. Among different types of procedures, MRI is the most trusted one to detect brain diseases. Manual and semi-automated segmentations need highly experienced radiologists and much time to detect the problem. Recently, deep learning methods have taken attention due to their automation and self-learning techniques. To get a faster result, we have used different algorithms of Convolutional Neural Network (CNN) with the help of transfer learning for classification to detect diseases. This procedure is fully automated, needs less involvement of highly experienced radiologists, and does not take much time to provide the result. We have implemented six deep learning algorithms, which are InceptionV3, ResNet152V2, MobileNetV2, Resnet50, EfficientNetB0, and DenseNet201 on two brain tumor datasets (both individually and manually combined) and one Alzheimer’s dataset. Our first brain tumor dataset (total of 7,023 images-training 5,712, testing 1,311) has 99-100 percent training accuracy and 98-99 percent testing accuracy. Our second tumor dataset (total of 3,264 images-training 2,870, testing 394) has 100 percent training accuracy and 69-81 percent testing accuracy. The combined dataset (total of 10,000 images-training 8,000, testing 2,000) has 99-100 percent training accuracy and 98-99 percent testing accuracy. Alzheimer’s dataset (total of 6,400 images-training 5,121, testing 1,279, 4 classes of images) has 99-100 percent training accuracy and 71-78 percent testing accuracy. CNN models are renowned for showing the best accuracy in a limited dataset, which we have observed in our models.

Author 1: Farhana Alam
Author 2: Farhana Chowdhury Tisha
Author 3: Sara Anisa Rahman
Author 4: Samia Sultana
Author 5: Md. Ahied Mahi Chowdhury
Author 6: Ahmed Wasif Reza
Author 7: Mohammad Shamsul Arefin

Keywords: Brain MRI; tumor; deep learning; classification; transfer learning

PDF

Paper 110: Toward A Holistic, Efficient, Stacking Ensemble Intrusion Detection System using a Real Cloud-based Dataset

Abstract: Network intrusion detection is a key step in securing today’s constantly developing networks. Various experiments have been put forward to propose new methods for resisting harmful cyber behaviors. Though, as cyber-attacks turn out to be more complex, the present methodologies fail to adequately solve the problem. Thus, network intrusion detection is now a significant decision-making challenge that requires an effective and intelligent approach. Various machine learning algorithms such as decision trees, neural networks, K nearest neighbor, logistic regression, support vector machine, and Naive Bayes have been utilized to detect anomalies in network traffic. However, such algorithms require adequate datasets to train and evaluate anomaly-based network intrusion detection systems. This paper presents a testbed that could be a model for building real-world datasets, as well as a newly generated dataset, derived from real network traffic, for intrusion detection. To utilize this real dataset, the paper also presents an ensemble intrusion detection model using a meta-classification approach enabled by stacked generalization to address the issue of detection accuracy and false alarm rate in intrusion detection systems.

Author 1: Ahmed M. Mahfouz
Author 2: Abdullah Abuhussein
Author 3: Faisal S. Alsubaei
Author 4: Sajjan G. Shiva

Keywords: Intrusion detection system; IDS dataset; stacking ensemble ids; stacking; security; ensemble learning

PDF

Paper 111: Authorship Attribution on Kannada Text using Bi-Directional LSTM Technique

Abstract: Author attribution is the field of deducing the author of an unknown textual source based on certain characteristics inherently present in the author’s style of writing. Author attribution has a ton of useful applications which help automate manual tasks. The proposed model is designed to predict the authorship of the Kannada text using a sequential neural network with Bi-Directional Long Short Term Memory layers, Dense layers, Activation function and Dropout layers. Based on the nature of the data, we have used stochastic gradient descent as an optimizer that improves the learning of the proposed model. The model extracts Part of the speech tags as one of the semantic features using the N-gram technique. A Conditional random fields model is developed to assign Part of the speech tags for the Kannada text tokens, which is the base for the proposed model. The parts of the speech model achieve an overall 90% and 91% F1 score and accuracy respectively. There is no state-of-art model to compare the performance of our model with other models developed for the Kannada language. The proposed model is evaluated using the One Versus Five (1 vs 5) method and overall accuracy of 77.8% is achieved.

Author 1: Chandrika C P
Author 2: Jagadish S Kallimani

Keywords: Authorship attribution; Bi-Directional Long Short Term Memory; machine learning algorithms; parts of speech; stylometry features

PDF

Paper 112: Flood Prediction using Deep Learning Models

Abstract: Deep learning has recently appeared as one of the best reliable approaches for forecasting time series. Even though there are numerous data-driven models for flood prediction, most studies focus on prediction using a single flood variable. The creation of various data-driven models may require unfeasible computing resources when estimating multiple flood variables. Furthermore, the trends of several flood variables can only be revealed by analysing long-term historical observations, which conventional data-driven models do not adequately support. This study proposed a time series model with layer normalization and Leaky ReLU activation function in multivariable long-term short memory (LSTM), bidirectional long-term short memory (BI-LSTM) and deep recurrent neural network (DRNN). The proposed models were trained and evaluated by using the sensory historical data of river water level and rainfall in the east coast state of Malaysia. It were then, compared to the other six deep learning models. In terms of prediction accuracy, the experimental results also demonstrated that the deep recurrent neural network model with layer normalization and Leaky ReLU activation function performed better than other models.

Author 1: Muhammad Hafizi Mohd Ali
Author 2: Siti Azirah Asmai
Author 3: Z. Zainal Abidin
Author 4: Zuraida Abal Abas
Author 5: Nurul A. Emran

Keywords: Deep learning; recurrent neural network; long short-term memory; flood prediction; layer normalization

PDF

Paper 113: Recognition Method of Dim and Small Targets in SAR Images based on Machine Vision

Abstract: Aiming at the problems of long recognition time and low recognition accuracy of traditional SAR image dim target recognition methods, a method of SAR image dim target recognition based on machine vision was proposed. SAR images are collected and preprocessed by machine vision, and the image information is processed by PCA dimension reduction considering the linear characteristics of the data to extract image features. Then, the SAR image target feature key frame frequency band is divided by the segmentation results, and the recognition model is established based on the image trajectory tracking and target analysis. The proposed algorithm is applied and analyzed. The simulation results show that the proposed algorithm has good recognition rate, average recognition rate and false detection rate are 99% and 0.9%, and can effectively ensure the data processing performance.

Author 1: Qin Dong

Keywords: Machine vision; SAR image; Weak target; PCA linear dimensionality reduction method; key frame frequency band

PDF

Paper 114: Information Classification Algorithm based on Project-based Learning Data-driven and Stochastic Grid

Abstract: The adaptive partitioning algorithm of information set in simulation laboratory based on project-based learning data-driven and random grid is studied to effectively preprocess the information set and improve the adaptive partitioning effect of the information set. Using the improved fuzzy C-means clustering algorithm driven by project-based learning data, the fuzzy partition of information set in simulation laboratory is carried out to complete preprocessing of information set; The pre-processing information set space is roughly divided by the grid partitioning algorithm based on the data histogram; A random mesh generation algorithm based on uniformity is used to finely divide the coarse mesh cells; Taking the representative points of grid cells as the clustering center, the pre-processing information set is clustered by the density peak clustering algorithm to complete the adaptive partitioning of the information set in simulation laboratory. Experimental results show that this algorithm can effectively preprocess and adaptively partition the information set of simulation laboratory; For different dimension information sets, the evaluation index values of Rand index, Purity, standard mutual information, interval and Dunn index of the algorithm are all high, and the evaluation index values of compactness and Davidson's banding index are all low, so the algorithm has a high accuracy of adaptive partitioning of information sets.

Author 1: Xiaomei Qin
Author 2: Wenlan Zhang

Keywords: Adaptive partitioning; data driven; information set; project-based learning; random grid; simulation laboratory

PDF

Paper 115: Swine flu Detection and Location using Machine Learning Techniques and GIS

Abstract: The H1N1 virus, more commonly referred to as swine flu, is an illness that is extremely infectious and can in some cases be fatal. Because of this, the lives of many individuals have been taken. The disease can be transmitted from pigs to people. This research presents an artificial neural network (ANN) classifier for disease forecasting, as well as a technique for detecting people who are sick based on the geographic region in which they are found. The source codes for these two algorithms are provided below. These coordinates serve as the foundation for the GIS coordinates that are utilized in the method for assessing the extent to which the illness has spread. The ICMR and NCDC datasets were utilized in the study. They used Dynamic Boundary Location algorithm to detect swine flu affected person’s location, the researchers discovered that the accuracy of the proposed classifier was 96 standard classifiers.

Author 1: P. Nagaraj
Author 2: A. V. Krishna Prasad
Author 3: V. B. Narsimha
Author 4: B. Sujatha

Keywords: Swine Flu; influenza; machine learning; GIS; classifiers; ANN; virus; algorithm

PDF

Paper 116: Taxation Transformation under the Influence of Industry 4.0

Abstract: Today the growing level of automation and the new concept of online technologies are transforming the traditional industry. Value is generated with the help of Industry 4.0 technologies that not only increase the efficiency and agility of supply chains, create new products and offer new ways of connecting businesses and consumers but also have a major impact on the traditional tax system. This study aims at determining changes in the modern industrial economy and substantiating possible directions for transforming the tax system to adapt it to the requirements of Industry 4.0. The objective is to identify the relationship between the digitalization of the economy, the use of blockchain technologies, robotics, automation, M2M technologies offered by Industry 4.0, and taxation. The article demonstrates how these technologies influence taxes and proposes measures to address possible tax issues. The authors of the article have concluded that the reasons (and goals) for transforming the current tax system as a result of the development of Industry 4.0 technologies are as follows: 1) to increase or stabilize tax revenues to compensate for tax losses and finance new education needs; 2) to introduce innovations for the development of Industry 4.0 and further digitalization of the economy; 3) to create an automatic tax administration system.

Author 1: Pavel Victorovich Stroev
Author 2: Rafael Valiakhmetovich Fattakhov
Author 3: Olga Vladimirovna Pivovarova
Author 4: Sergey Leonidovich Orlov
Author 5: Alena Stanislavovna Advokatova

Keywords: Digitalization; cryptocurrency; blockchain; robotics; automation; innovation; tax

PDF

Paper 117: Attractiveness of the Megaproject Labor Market for Metropolitan Residents in the Context of Digitalization and the Long-Lasting COVID-19 Pandemic

Abstract: The article aims to determine the nature of changes in the attractiveness of the labor market of megaprojects from the perspective of megapolis residents under the conditions of digitalization and the long-lasting pandemic of COVID-19. The paper develops a scientific-methodological and categorical-conceptual apparatus with the support of empirical methods with distance methods. The study shows that the attractiveness of the labor market of megaprojects has undergone certain changes for megapolis residents under the current conditions. The factors of the attractiveness of the labor market of megaprojects are of a stable nature in the minds of megapolis residents. The main advantage of the work is the identification of trends in the changes of the megaproject labor market and the relationships they have. The study reveals both general and private trends. The obtained results can be used for further study of the megaproject labor market and the improvement of the social policy of the state and megalopolises in the conditions of digitalization and the prolonged pandemic.

Author 1: Mikhail Vinichenko
Author 2: Sergey Barkov
Author 3: Aleksander Oseev
Author 4: Sergey Makushkin
Author 5: Larisa Amozova

Keywords: Megaproject labor market; metropolitan residents; digitalization; COVID-19 pandemic; attractiveness factors

PDF

Paper 118: Generation and Assessment of Intellectual and Informational Capital as a Foundation for Corporations’ Digital Innovations in the “Open Innovation” System

Abstract: The research peruses development of scientifically based toolkit to create and assess promising types of intellectual capital transformed into digital innovations for “open innovation" system. It is determined that in theory terms “intellectual, informational and digital capitals” are interrelated categories; efficient merge of informational and digital capitals minimizes information security risks; merge of informational and digital capitals provides a long-term multiplicative synergetic effect demonstrating constant transformation of innovative ideas into digital innovations. The following is suggested: structural and logical scheme method for creation and assessment of informational capital and scenarios for the synergetic development of informational and digital capitals.

Author 1: Viktoriya Valeryevna Manuylenko
Author 2: Galina Alexandrovna Ermakova
Author 3: Natalia Vladimirovna Gryzunova
Author 4: Mariya Nikolaevna Koniagina
Author 5: Alexander Vladimirovich Milenkov
Author 6: Liubov Alexandrovna Setchenkova
Author 7: Irina Ivanovna Ochkolda

Keywords: Informational and digital capitals; informational security risks; synergy; open innovation; transformation

PDF

Paper 119: An Algorithm for Providing Adaptive Behavior to Humanoid Robot in Oral Assessment

Abstract: Assistance humanoid robots (AHR) are the category of robotics used to offer social interaction to humans. In higher education, the teaching staff supports the acceptance of AHRs as a social assistance tool during the learning activities, with the whole responsibility of the correct operation of the device and providing a more comprehensive view of the objectives and significance of AHR use. On the other hand, students deal with AHRs either as a friend or control figures as a teacher. This paper presents an algorithm for AHRs in oral assessments. The proposed algorithm focuses on four characteristics: adaptive occurrence, friendly existence, persuasion, and external appearance. This paper integrates AHRs in higher education to improve the value of psychological and social communication during oral assessment where can assist students in dealing with challenges, such as shyness, dissatisfaction, hesitation, and confidence, better than a human teacher can. Thus, AHRs have increased students’ self-confidence and enriches active learning.

Author 1: Dalia khairy
Author 2: Salem Alkhalaf
Author 3: M. F. Areed
Author 4: Mohamed A. Amasha
Author 5: Rania A. Abougalala

Keywords: Algorithm; humanoid robot; social robots; oral assessment; assistance robots; higher education; adaptive behavior robot

PDF

Paper 120: Classifiers Combination for Efficient Masked Face Recognition

Abstract: This study was developed following the upheaval caused by the spread of the Coronavirus around the world. This global crisis greatly affects security systems based on facial recognition given the obligation to wear a mask. This latter, camouflages the entire lower part of the face, which is therefore a great source of information for the recognition operation. In this article, we have implemented three different pre-trained feature extractor models. These models have been improved by implementing the well-known Support Vector Machines (SVM) to reinforce the classification task. Among the investigated architectures, the FaceNet feature extraction model shows remarkable results on both databases with a recognition rate equal to 90%on RMFD and a little lower on SMFD with 88.57%. Following these simulations, we have proposed a combination of classifiers (SVM-KNN) that would prove a remarkable improvement and a significant increase in the accuracy rate of the selected model with almost 4%.

Author 1: Kebir Marwa
Author 2: Ouni Kais

Keywords: Masked faces; deep learning; AlexNet; ResNet50; FaceNet; classifiers combination

PDF

Paper 121: Human Position and Object Motion based Spatio-Temporal Analysis for the Recognition of Human Shopping Actions

Abstract: Retailers have long sought ways to better understand their consumers' behavior in order to deliver a smooth and enjoyable shopping experience that draws more customers every day and, as a result, optimizes income. By combining various visual clues such as activities, gestures, and facial expressions, humans may fully grasp the behavior of others. However, due to inherent problems as well as extrinsic forced issues such as a shortage of publicly available information and unique environmental variables, empowering computer vision systems to provide it remains an ongoing problem (wild). In this paper, the authors focus on identifying human activity recognition in computer vision, which is the first and by far the most important cue in behavior analysis. To accomplish this, the authors present an approach by integrating human position and object motion in order to detect and classify tasks in both temporal and spatial analysis. On the MERL shopping dataset, the authors get state-of-the-art results and demonstrate the capabilities of the proposed technique.

Author 1: Nethravathi P. S
Author 2: Karuna Pandith
Author 3: Manjula Sanjay Koti
Author 4: Rajermani Thinakaran
Author 5: Sumathi Pawar

Keywords: Deep convolutional neural networks; computer vision; object detection; object localization; temporal analysis; human shopping actions component

PDF

The Science and Information (SAI) Organization
BACK TO TOP

Computer Science Journal

  • About the Journal
  • Call for Papers
  • Submit Paper
  • Indexing

Our Conferences

  • Computing Conference
  • Intelligent Systems Conference
  • Future Technologies Conference
  • Communication Conference

Help & Support

  • Contact Us
  • About Us
  • Terms and Conditions
  • Privacy Policy

© The Science and Information (SAI) Organization Limited. All rights reserved. Registered in England and Wales. Company Number 8933205. thesai.org