Research On Network Intrusion Detection Based On Heuristic Feature Search

In recent years,as one of the most important infrastructure,the Internet is facing increasingly serious security problems.The network intrusion behavior is complex and changeable.The strategy of collecting attack behavior patterns and adding defense rules is far from enough to resist risks.Therefore,the research focused on network intrusion detection and defense gradually turns to the algorithm based on heuristic significance.With the rise of machine learning and data mining technology,intrusion detection based on the characteristics of network traffic behavior has become an important research direction in the field of network security.In this paper,based on the improved and further optimized genetic algorithm and neural network algorithm,an intrusion detection model is established.This paper focuses on the process of heuristic feature search and selection of the topological structure of the neural network model by using the optimized genetic algorithm.The neural network algorithm is used to mine the effective feature information in the network connection message,and the genetic algorithm is used to optimize the neural network model,and optimize the data feature search mechanism.Finally,the different versions of genetic algorithm are compared and statistically analyzed.The results show that the optimized genetic algorithm is highly compatible with the neural network model,and can get excellent results in the detection of KDD99 network intrusion data set.At the same time,the results are compared with the results of support vector machine,ant colony algorithm and different versions of genetic algorithm.The results show that the improved intrusion detection model in this paper has higher accuracy and lower false-positive and false negative rates in the test of multiple data sets.The main content of this paper is divided into the following 4 parts.(1)Based on the genetic algorithm,the multi-faceted detail optimization is carried out,and the cross and mutation generations with stronger search ability in the parameter space are constructed.The dynamic design of adaptive adjustment of super parameters with the iterative process is added,and a new fitness function is proposed to be compatible with the training effect of the neural network model.(2)Based on the neural network algorithm to optimize the structure scale,add the on-off control mechanism between the neuron node and the node connection,ensure theeffective node and the connection to remain active while eliminating the redundant structure,in order to maintain the classification and discrimination ability of the data set as small as possible.(3)In six different benchmark functions,different versions of genetic algorithm and neural network models,including the optimization algorithm in this paper,are tested,and the accuracy and model size are compared.(4)Design the core framework of network intrusion detection model,test the improved and fused genetic algorithm and neural network model on KDD99 network intrusion data set,and compare with other mainstream heuristic algorithms.