Abstract
In this paper, an intelligent and lightweight anomalous network traffic detection framework is proposed. The framework uses principal component analysis (PCA) with the main purpose of feature extraction and dimensionality reduction and bidirectional generative adversarial network (BiGAN) model is used to detect the anomalous network traffic. The proposed framework was evaluated using KDDCUP-99 dataset and was compared with recent deep learning models. Various visualization methods were also employed to understand the characteristics of the dataset and to visualize the KDDCUP-99 dataset features. Our work shows the importance of feature reduction in improving the overall performance of the BiGAN models. In addition, the proposed models are faster and efficient at test time.