an improved intrusion detection system utilizing a new mix of bp and som neural networks

high processing loads, need for complicated and frequent updating, and high false alarm are some of the challenges in designing anomaly detection and misuse detection systems. we propose a new network-based intrusion detection system (ids) that resolves such shortcomings. our scheme fuses anomaly detection and misuse detection systems, which has not been utilized so far in existing systems. in doing so, we have employed a mix of modified back-propagation (bp) and self-organizing map (som) neural networks that perform pattern recognition and classification in an effective and efficient manner. results indicate that the performance of our proposed ids is significantly improved as compared to the existing systems.