Research On Intrusion Detection Method Based On Convolutional Neural Network For Provenance Diagram

Nowadays,Internet technology is advancing rapidly,and information security issues are gradually entering people's horizons.The diversification of intrusion methods and the explosive growth of stored data make traditional intrusion detection methods unable to meet all the requirements of current network security.At present,the traceability-based intrusion detection method has an excessive amount of stored data,a large time overhead,and a low detection accuracy.In view of the detection effect and space overhead of the current intrusion detection technology,it is proposed to combine the convolutional neural network technology with the traceable intrusion detection technology to improve the usability of the detection method.Aiming at the situation that the data volume is too large,a method of transforming the traceability map information into feature vectors is proposed.This method selects the center node by assigning the node importance degree to each node in the traceability map information,and constructs a neighborhood for the central node to generate a neighborhood.The vector information that can be processed by the convolutional neural network reduces the storage space of the detection data set.For the problem of low detection accuracy,a convolutional neural network model applied to the traceable data is constructed,and the dependence between nodes in the traceability map is extracted.Feature information such as relationships and attributes to predict unknown attacks,making the detection results more accurate.The experimental results show that the detection method proposed in the paper is 1%?37% higher than the intrusion detection system based on traceability and traceability path.The detection time is shortened by 5%?25%,and the storage space is reduced by 1%?24%;Compared with the traceability-based real-time misuse detection algorithm,the detection accuracy is improved by 32%?72%,the detection time is shortened by 80%?99%,and the storage space is reduced by 87%?99%.