Intrusion detection method based on improved BP neural network algorithm

The BP neural network algorithm is improved to detect intrusion. The improved method is as follows: according to different input, different part of network will be fired to generate output. The weights of the arcs only connected with fired neuron will be adjusted whereas all weights in traditional BP network must be adjusted. The experiment results indicate that this improved algorithm can increase learning speed and shorten training time. The results also show that the intrusion detection rate is improved.