Cyber Hacking Breaches Prediction using Machine Learning

Abstract

Cyber-attacks are a major threat to these systems. Unlike faults that occur by accidents in cyber-physical systems, cyber-attacks occur intelligently and stealthily. Some of these attacks which are called deception attacks, inject false data from sensors or controllers, and also by compromising with some cyber components, corrupt data, or enter misinformation into the system. If the system is unaware of the existence of these attacks, it won’t be able to detect them, and performance may be disrupted or disabled altogether. The proposed method in this study is to use the structure of deep neural networks for the detection phase, which should inform the system of the existence of the attack in the initial moments of the attack. In the presented control method, after the attack detection phase with the use of a deep neural network, the control system uses the reputation algorithm to isolate the misbehaving agent. Experimental analysis shows us that deep learning algorithms can detect attacks with higher performance than usual methods and can make cyber security simpler, more proactive, less expensive, and far more effective