An Intrusion Detection Algorithm of Dynamic Recursive Deep Belief Networks
Jingjing Tian, Pingan Li · 2017
In order to improve the accuracy of intrusion detection,the algorithm of dynamic recursive deep belief networks based on random forest is proposed in this paper. First of all, deep belief networks are constructed by the analysis of the deep belief networks characteristics and abstract features which are suitable for classification are extracted. Then we introduce the tracking time-varying factor and the forgetting coefficient to make the parameters we have trained more reasonable. Finally, the features we have extracted are classified by random forest. Through the experiment, this method can ensure higher detection rate and lower false rate than other algorithms. This method proves to be effective.