A survey on types of machine learning techniques in intrusion prevention systems

The computation technology is evolving. The data transmitted and generated using them are growing exponentially. The traffic on these networks requires surveillance. Effective network traffic surveillance, packet analysis and rules to define the traffic flow are in place using various intrusion detection and prevention systems. However, the security issues and concerns are dictating the need to evolve with the novel and `human-like' methods to mitigate them. Normal methods and techniques may prove too tedious, and sometimes will not even render fruitful results. In that case, the techniques that can think, act and mimic human beings assure the possibilities of resolving the concerns where the domain of artificial intelligence plays a critical role. Also, there is a crucial need to design real-time models that can analyze & handle bigger, more complex data; and deliver faster as well as accurate results. Machine learning is a subfield which possesses all of the attributes. Intrusion preventions IDPS's integrate methods that help to prevent intrusive and non-intrusive data packets [2]. According to survey, most of the intrusion prevention methods rely heavily on humans to analyze data and classify into intrusive and non-intrusive networks. When the data to compute is very large, computing methods such as Computational intelligence and its constituent are essential to use as humans have their own shortcomings. Computational intelligence and machine learning are widely used to detect the intrusions, which in turn help in preventing the same. Pattern recognition uses algorithms to handle very large data sets. In this paper, a survey is done that on various machine learning and computational intelligence techniques that envisage using IDPS.