Enhancing detection rates in intrusion detection systems using fuzzy integration and computational intelligence

Abstract

Intrusion Detection Systems (IDS) show a major part in computer cyber defense by detecting and reacting to unauthorized activities. These systems monitor network and system activity, evaluating developments to identify possible security breaches. Enhancing Detection Rates in IDS includes optimizing algorithms, employing Machine Learning (ML) approaches, and employing intrusion detection to enhance the system's functionality to find novel vulnerabilities immediately. Continuous improvement in detection capabilities is essential for adapting to evolving challenges from cyberspace and maintaining resilience of the online infrastructure. To enhance the detection rates, data preprocessing like min-max normalization, followed by t-distributed Stochastic Neighbor Embedding (t-SNE) feature extraction technique to capture most discriminative attributes for attack classifications. The established Genetic Fuzzy Systems (GFS) throughout paired learning framework for detecting input attack. The model enhances accuracy for unusual attack occurrences by better distinguishing between normal activity and distinct attack categories. To proposed Generative Adversarial Network (GAN) as a classifier for enhancing detection rates. This research explores the performance of the proposed GFS-GAN model on two prominent intrusion detection datasets are the TII-SSRC-23 for dataset 1 and NSL-KDD for dataset 2. The suggested GFS-GAN model demonstrated exceptional performance on the TII-SSRC-23 dataset, achieving 99.23 % accuracy. The GFS-GAN model also performed well on the NSL-KDD dataset, with an accuracy of 99.13 %, The findings illustrate GANs' capabilities to progress the efficacy and durability of IDS, resulting in effective protection against complicated cyber-attacks.