基于BiLSTM和注意力机制的入侵检测

Intrusion detection based on BiLSTM and attention mechanism

由于传统网络入侵检测方法的局限性无法满足当前网络安全需要,为提高网络入侵检测的准确率,引入机器学习算法,提出一种基于BiLSTM(bi-directional long short-term memory)和注意力机制的网络入侵检测方法。利用BiLSTM网络进行长距离依赖特征提取,利用DNN(deep neural network)提取更深层次的特征,引入注意力机制增加对特征重要性的计算,通过softmax分类器获得分类结果。使用NSL_KDD数据集作为实验数据,实验结果表明,相比于对比方法,该方法有效提高了入侵检测的准确率,验证了该方法的有效性。

Because the limitations of traditional network intrusion detection methods can not meet the need of current network security,a network intrusion detection method based on BiLSTM(bi-directional long short-term memory)and attention mechanism was proposed to improve the accuracy of detection by introducing machine learning algorithm.BiLSTM network was used to extract long distance dependent features,DNN(deep neural network)was used to extract deeper features,and attention mechanism was used to increase the calculation of feature importance.The classification results were obtained by using softmax classifier.NSL_KDD data set was used as experimental data.The experimental results show that compared with the contrast method,the accuracy of intrusion detection is improved,and the effectiveness of the method is verified.