Intelligent Feature Selection Model using Artificial Neural Networks for Independent Cyberattack Classification
Lakshmi Chandrakanth Kasireddy, Hari Prasad Bhupathi, Rohit Shrivastava, Prem Kumar Sholapurapu, Nirav Bhatt, Ratnamala · 2025
To enhance the ability of cyber-attack detection, this research proposes an ANN-INDFSM. Since there has been an increase in technology, IDS play an important role in protecting networks by detecting intruders. However, conventional IDS models encounter problems because of the massive amount of data considered for analysis in which the feature selection plays crucial role in terms of accuracy and computation time period. ANN-INDFSM minimizes computational complexity while achieving higher detection accuracy due to the consideration of the non-dependent features only. This model improves the prediction by using data preprocessing techniques, feature extraction, and weight assignment and performs at 97.58 % of accuracy in intrusion detection. In contrast to earlier models, which may actually be hampered by the presence of more data, ANN-INDFSM incorporates feature selection through the use of weights, which minimizes false alarms besides enhancing the speed of classification. Possible future improvements are then the combination with other approaches, for example hybrid optimization, and the exploitation of larger datasets during training in order to improve detection. This approach shows much potential of improving IDS capability against emerging cyber threats.