Multi-Classifier Deep Neural Network for Detecting Intruder Behavior In Cyber Security
Desire Iradukunda, Xiong Wan An, Waqar Ali · 2019
The security is becoming much more important due to the massive growth in computer network technologies. Finding anomalies from a system's network traffic has been a hot research topic over the last two decades. Most of the existing signature- based intrusion detection systems only rely upon the pre- configured and predetermined attack patterns that are practically not true for real life challenges. In this research work, we attempt to capture the dynamic behavior of intruders by utilizing a novel Multi-Classifier Deep Neural Network (MCDNN) framework. The proposed framework intuitively segregate the attackers and authorize user data packets for a given network traffic. The MCDNN is based on four layers architecture, each layer is based upon the logistic regression classifier and is fully connected. The MCDNN utilizes a two- stage novel feature learning framework. The model is capable to automatically learn useful features from large network traffic labeled data collection and classifies intruders efficiently. We evaluate the effectiveness of our proposed model on a well- known publicly available challenging dataset KDD99. The experimental results demonstrate that our proposed framework significantly outperforms the existing models and achieves considerable high recognition rates on KDD99.