Intrusion Detection Algorithm Based on CNN and LSTM

Aiming at the shortcoming that traditional intrusion detection algorithms ignore the time series characteristics of data sets, a proposed intrusion detection algorithm integrates a two-branch convolutional neural network with long short-term memory capabilities. The convolutional neural network employs varying-sized convolutional filters to discern diverse hierarchical features within the data. Additionally, it employs a collaborative energizing mechanism to enhance the model's universal applicability. The long short-term memory is used to extract the time series characteristics of traffic data, and Bayesian algorithm is used to optimize the selection of super parameters to solve the problem of model determination. The algorithm is verified on CICIDS2017 data set and compared with other algorithms. The results show that the algorithm has higher accuracy and precision, with the accuracy rate reaching 98.85% and the precision reaching 98.12%.