Intrusion detection method research based on optimized self-buildup clustering neural network

This paper puts forward a method of bringing neural network to bear intrusion detection. When the average error can´t decrease any longer, the hereditary algorithm will be used to continuatively train the network in the interest of acquiring optimized join parameter. The network structure and network joining parameter will evolve at the same time by the neural network and hereditary algorithm. The convergence effect is good and the adaptivity is strong, suitable for real-time processing.