RESNETCNN: An abnormal network traffic flows detection model

Intrusion detection is an important means to protect system security by detecting intrusions or intrusion attempts on the through operational behaviors, logs, and data audit. However, existing systems suffer from incomplete feature extraction low classification accuracy, which affects effect. To this end, paper proposes model that fuses residual network (RESNET) parallel crossconvolutional neural network, called RESNETCCN. RESNETCNN can efficiently learn various stream features fusion of deep learning convolutional (CNN), improves accuracy abnormal streams in unbalanced streams, moreover, oversampling method into preprocessing, extract multiple types at same time, effectively solving problems streams. Finally, three improved versions networks are designed meet requirements different traffic processing, highest reaches 99.98% CICIDS 2017 dataset 99.90% ISCXIDS 2012 dataset.