Design of Computer Network Security Defense System Based on Artificial Intelligence and Neural Network

In order to improve the effect of computer security defense, this paper applies artificial intelligence technology and neural network technology to the design of computer network security defense system, and proposes the simplification theorem of softmax function and rigorously proves its correctness. The theorem points out that under the condition that a certain accuracy of the output probability is guaranteed, the calculation efficiency of the softmax function can be improved by omitting a large number of exponential operations. In addition, in order to deal with the flow table overload attack in SDN, this paper merges the unused flow table space of all switches in the entire network. Through the experimental research results, it can be seen that the computer network security defense system based on artificial intelligence and neural network proposed in this paper has a good security defense effect.