As cyber attacks become more advanced and more frequent, it is becoming infeasible to develop secure systems with human intervention and static software alone. As a result, researchers are developing artificial intelligence (AI) techniques to defend against cyber attacks. Artificial intelligence has many benefits over humans; it is “flexible, adaptable and robust, and able to detect a wide variety of threats and make intelligent real-time decisions. “They are often built into intrusion detection and protection systems (IDPS). These are pieces of software or hardware that protect the network by detecting attacks and preventing them. Current attempts at creating artificial intelligence defense systems can be broken into three classes: Artificial neural networks, intelligent agents, and artificial immune systems.
Artificial neural networks simulate neural networks in biological systems, and are good at tasks like prediction and classification. This makes them good tools for tasks such as intrusion detection and attack classification. While neural networks are good at classifying and predicting, they are not good at performing tasks themselves. Intelligent agents are autonomous programs that are able to communicate with each other and execute responses to foreign attacks. The paper further goes on to describe two more subfields of artificial intelligence. They describe artificial immune systems which are self adaptive and self organizing, and genetic algorithms which give optimal solutions in complex environments. This paper offers an in depth survey of these methods, and the use of artificial intelligence for cyber security.
