Combining Feature Dimensionality Reduction and NGO-CNN-BiLSTM′s Network Abnormal Traffic Detection Method in Recruitment Intelligence Platform

In view of the challenges brought by computer viruses and network malicious attacks to the operation and maintenance of the recruitment intelligent platform system, this paper proposes a method to detect abnormal traffic of the recruitment smart platform based on feature dimensionality reduction combined with Northern Goshawk Optimization ( NGO) to optimize convolutional neural network (CNN) and Bi-directional Long Short-Term Memory (BiLSTM). By using the Kernel Principal Component Analysis (KPCA) method to reduce the dimensionality of the nonlinear information contained in the dataset, the reduced data is used as input for the deep learning network model. Then, the convolution kernel of CNN is optimized by NGO to obtain the optimal convolution kernel, and the abnormal traffic was detected by BiLSTM. The CIC-IDS-2017 data set is used to analyze the training and test samples of the intrusion detection network model, and the accuracy and training time are improved compared with other methods, which confirms the feasibility and effectiveness of this method.