Network intrusion detection based on conditional wasserstein variational autoencoder with generative adversarial network and one-dimensional convolutional neural networks

Abstract There is a class-imbalance problem that the number of minority class samples significantly lower than majority in common network traffic datasets. Class-imbalance phenomenon will affect performance classifier and reduce robustness to detect unknown anomaly detection. And distribution continuous features dataset does not follow Gaussian distribution, which bring great difficulties intrusion We propose Conditional Wasserstein Variational Autoencoders with Generative Adversarial Network (CWVAEGAN) solve phenomenon, CWVAEGAN transform original through data preprocessing, then use improved VAEGAN generate samples. According model, an detection system based on One-dimensional convolutional neural networks (1DCNN), namely CWVAEGAN-1DCNN, established. By using examples generated by CWVAEGAN, unbalanced solved. Specifically, CWVAEGAN-1DCNN consists three modules: preprocessing module, deep network. evaluate two benchmark datasets compared it other 16 methods. Experiment results suggest better class-balancing methods, advanced