Research on Intrusion Detection Model Based on DBN and Random Forest

• 1、School of Cyberspace Security, Beijing University of Posts and Telecommunications, Beijing 100876

Abstract：Aiming at the problems of traditional machine learning algorithms in the field of intrusion detection, such as poor computing performance, poor classification results, and poor generalization performance, as well as the current large-scale network traffic and its complexity. this paper proposes a method based on deep belief networks (DBN) To perform feature learning, and use an ensemble learning method such as random forest to construct an intrusion detection model for the classifier. Aiming at the complexity of network traffic, multi-layer restricted Boltzmann machines are used for feature learning to achieve the purpose of dimensionality reduction; random forest is used to classify and identify data after dimensionality reduction to improve detection accuracy and generalization of intrusion detection models. performance. Experiments with the KDD99 dataset show that the intrusion detection model combining DBN and random forest is efficient and feasible, and has higher detection accuracy.

Keywords： Intrusion detection deep learning ensemble learning deep belief network (DBN) dimensionality reduction random forest