ABSTRACT To address the problems of low accuracy and high false alarm rate of traditional distributed intrusion detection methods, a distributed intrusion node detection algorithm based on an immune neural network is proposed. The motivation for the research stems from the increasing complexity of cybersecurity threats and the limitations of existing technologies in dealing with new types of attacks. Through experimental contrast learning algorithm and machine learning tool, the experimental results indicated that the proposed algorithm was superior to existing methods in key performance indicators such as recognition rate, false positive rate, and false negative rate. The specific recognition rate reached 96 %, the false positive rate was only 4 %, and the false negative rate dropped to 3 %. This finding not only validated the effectiveness of the proposed algorithm but also provided theoretical support and practical value for future network security research. The main contribution is to propose an innovative hybrid model that successfully combines biological immune mechanisms with deep learning to significantly improve the performance and adaptability of intrusion detection systems.
Tianzhu Guan (Tue,) studied this question.