Dynamic Ant Colony Optimization–Driven Feature Selection for GRU-Based Network Intrusion Detection

Modern network environments generate large-scale, high-dimensional traffic data that can be effectively interpreted as complex temporal signals. Efficiently analyzing such data requires robust feature selection and accurate temporal modeling techniques. In this study, we propose a signal-driven intrusion detection framework that integrates a Dynamic Ant Colony Optimization (DyACO) algorithm with deep recurrent neural modeling. The proposed DyACO performs adaptive wrapper-based feature selection by dynamically adjusting pheromone evaporation, heuristic guidance, and reinforcement strategies to enhance exploration in high-dimensional signal spaces. Selected features are then processed using a gated recurrent unit network to capture temporal dependencies in network traffic signals. The experiments on the CSE-CIC-IDS2018, UNSW-NB15, and NSL-KDD datasets demonstrate that our proposed method not only effectively compresses the feature space but also enhances classification accuracy, precision, recall, and F-score. The results demonstrate that when applied to network traffic, structured temporal signals, along with adaptive optimization modeling, yield an effective solution for intelligent, signal-based anomaly detection.

@article{Yang2026,
  title     = {Dynamic Ant Colony Optimization–Driven Feature Selection for GRU-Based Network Intrusion Detection},
  journal   = {International Journal of Advanced Computer Science and Applications},
  volume    = {17},
  number    = {9},
  year      = {2026},
  publisher = {The Science and Information Organization},
  author    = {Zhaosheng Yang},
  doi       = {10.14569/IJACSA.2026.0170959},
  url       = {https://doi.org/10.14569/IJACSA.2026.0170959}
}