Detection of Real Time Malicious Intrusions Using GAN (Generative Adversarial Networks) in Cyber Physical System
Sanjay Reddy Gunnam, Sameer Kumar Vepuri, V Nallarasan · 2024
The project understands that computer viruses, malware, and aggressive strikes on computer networks are a constant danger. This shows how important breach detection is as a proactive defense technology. Deep learning is used to detect and correct cyber security gaps and breaches in IoT-driven cyber-physical systems to improve security. The goal of the project is to make intruder detection better than what current systems can do by focusing on things like accuracy, efficiency, and lowering false positives. This draws attention to the progress and new ideas in defense. A generative adversarial network, a cutting-edge deep learning technique, is used by the method to reach the project's goals. It also stands out by comparing uncontrolled and deep learning-based methods to discrimination, showing a complete and effective way to handle privacy. In our project, we successfully used an ensemble method to improve the accuracy of our predictions by combining several separate models. It's worth mentioning that it has a mixed design that combines CNN and LSTM, which stands for Congressional Neural Networks and Long Short-Term Memory. When applied to the KDD-Cup dataset, this hybrid model got an amazing accuracy of 99%. This shows that our ensemble method works well for finding intrusions in IoT-based defense systems.