An Improved Deep Learning Based Intrusion Detection Method
Pan Wang, Xuehua Song, Zhuanglai Deng, Huimin Xie, Changda Wang · 2019
In order to solve the problem that the traditional intrusion detection method is difficult to accurately extract the feature information of massive unlabeled network data to identify the abnormal intrusion, this paper proposes an intrusion detection method of Softmax classification based on the improved deep belief network (IDBN-SC), where the deep belief network is employed for conduct unsupervised feature learning. The adaptive learning speed is applied to reduce the time required to reconstruct the error when training the network model. The pre-training technique is utilized to reduce the feature of the original intrusion data, and the back propagation algorithm is applied to obtain the optimal low-dimensional representation. Moreover, the improved Softmax classifier is utilized to identify the anomaly network intrusion. The experimental data shows that the IDBN-SC approach has higher detection accuracy and better real-time performance compared to the other methods such as IDBN-based support vector machine and IDBN-based original Softmax regression when using the same NSL-KDD dataset.