A Comprehensive Survey of Deep Learning-Based Adversarial Network Intrusion Detection Systems

With the growing complexity of cyber-attacks network intrusion detection system based on deep learning are more vulnerable to adversarial attacks, which alter input data to bypass detection. This study provides a thorough analysis of deep learning-based adversarial network intrusion detection systems, emphasizing the many adversarial attack types that target NIDS, generation algorithm (FGSM, JSMA, PGD, C&W. To improve the resilience and dependability of deep learning models, this study also investigates a variety of defense strategies, including defensive distillation and adversarial training. The summarized researcher's contribution by reviewing multiple datasets, assessing their pros and cons is presented, while proposing future pathways to mitigate emerging threats. This study provides a thorough analysis of adversarial attacks and suggests exciting avenues for making robust, secure intrusion detection systems.