A SRC-RF and WGANs-Based Hybrid Approach for Intrusion Detection

In contemporary network security environment, the Intrusion Detection System (IDS) stands as a pivotal safeguard against unauthorized activities. However, the effectiveness of IDS in accurately categorizing minority classes is often hampered by the challenges posed by imbalanced class issues inherent in the dataset. Consequently, devising a data balancing algorithm for unbalanced data becomes a complex task. In this paper, we propose an innovative approach to solve the imbalanced learning problem, which aims to balance the normal samples and attack samples. Firstly, we utilize WGAN to generate additional data for the processed dataset. This step has effectively augmented the minority class sample count, consequently establishing a more equitable and varied training ensemble. Then, we design SFC-RF to extract highly correlated features from the mixed dataset, which includes both the original and the generated data. The SRC-RF algorithm is committed to optimizing the feature selection process, aiming to enhance the model's discriminative ability and thereby improve its classification accuracy. Finally, a Convolutional Neural Network (CNN) is employed to leverage the features extracted by SRC-RF from the mixed dataset. The CNN is trained on this enriched and balanced dataset, learning to classify the instances accurately based on the discriminative features extracted by SRC-RF. The experimental results show that our proposed method outperforms the state-of-the-art methods and achieves better performance.