Imbalance Datasets in Malware Detection: A Review of Current Solutions and Future Directions

Imbalanced datasets are a significant challenge in the field of malware detection. The uneven distribution of malware and benign samples is a challenge for modern machine learning based detection systems, as it creates biased models and poor detection rates for malicious software. This paper provides a systematic review of existing approaches for dealing with imbalanced datasets in malware detection such as data-level, algorithm-level, and ensemble methods. We explore different techniques including Synthetic Minority Oversampling Technique, deep learning techniques including CNN and LSTM hybrids, Genetic Programming for feature selection, and Federated Learning. Furthermore, we assesses the strengths, weakness, and areas of application of each approach. Computational complexity, scalability, and the practical applicability of these techniques remains as challenges. Finally, the paper summarizes promising directions for future research like lightweight models and advanced sampling strategies to further improve the robustness and practicality of malware detection systems in dynamic environments.

@article{Almajed2025,
  title     = {Imbalance Datasets in Malware Detection: A Review of Current Solutions and Future Directions},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {16},
  number    = {1},
  year      = {2025},
  publisher = {The Science and Information Organization},
  author    = {Hussain Almajed and Abdulrahman Alsaqer and Mounir Frikha},
  doi       = {10.14569/IJACSA.2025.01601126},
  url       = {https://doi.org/10.14569/IJACSA.2025.01601126}
}