基于Adam-BNDNN的网络入侵检测模型

针对传统入侵检测算法检测精度低、误报率高等问题，提出了一种融合批量规范化和深度神经网络的网络入侵检测模型。该模型首先在深度神经网络隐藏层添加批量规范化层，优化隐藏层的输出结果，然后采用Adam自适应梯度下降优化算法对BNDNN参数进行自动优化，提高模型检测能力。并使用NSL-KDD数据集进行仿真实验，结果表明该模型的检测效果优于SNN、KNN、DNN等检测方法；整体检测率可达99.41%，整体误报率为0.59%，证明了模型的可行性。

The Network intrusion detection model based on Adam-BNDNN

Aiming at the problems of low detection accuracy and high false alarm rate of traditional intrusion detection algorithm, a network intrusion detection model combining batch normalization and deep neural network is proposed. Firstly, a batch normalization layer is added to the hidden layer of the deep neural network to optimize the output of the hidden layer, and then the adaptive gradient descent optimization algorithm of Adam is used to optimize the parameters of BNDNN automatically to improve the detection ability of the model. The simulation experiment with NSL-KDD data set shows that the detection effect of the model is better than shallow neural network (SNN), k-NearestNeighbor (KNN), deep neural network(DNN) and other detection methods; The overall detection rate is 99.41%, and the overall false alarm rate is 0.59%, which proves the feasibility of the model.