A comparative analysis of CGAN‐based oversampling for anomaly detection

In this work, the problem of anomaly detection in imbalanced datasets, framed context network intrusion is studied. A novel solution that takes both data-level and algorithm-level approaches into account to cope with class-imbalance proposed. This integrates auto-learning ability Reinforcement Learning oversampling a Conditional Generative Adversarial Network (CGAN). To further investigate potential CGAN, classification tasks, effect CGAN-based on following classifiers examined: Naïve Bayes, Multilayer Perceptron, Random Forest Logistic Regression. Through experimental results, authors demonstrate improved performance from proposed approach, general, over other techniques such as Synthetic Minority Oversampling Technique Adaptive Synthetic.