International Research Journal on Advanced Science Hub

With the advancement in the artificial intelligence technologies and develop- ment of fifth generation networks, a network may face many hazards and chal- lenges as the number of users are accessing the network simultaneously which makes the user to think of losing the confidentiality of the data and hence the network to be considered for security. Threats on the network can be classified in many ways and to detect such threats an Intrusion detection system (IDS) is the one which is mainly used. A network can be attacked in two ways as minor attack and major attack. Denial-of-Service (DoS) and Prob attacks belong to major kind and User-to-Root (U2R) and Remote-to-Login (R2L) goes to minor attack categories. The minor attacks are also called as rare attacks which are very injurious for a host and it is very difficult to recognize these attacks. This paper consists of a survey made on IDS and different algorithms used to imple- ment these IDSs using machine learning.

This paper focuses on preventing cyber attacks, which are common on any device connected to the internet. In order to create an intrusion detection system (IDS) that can recognise and differentiate cyber-attacks at the network and host levels in a timely and automated manner, machine learning techniques are widely used. A deep neural network (DNN) is a form of deep learning model being researched for use in developing a scalable and efficient intrusion detection system (IDS) capable of detecting and classifying unexpected and unpredictable cyber-attacks.Since network behaviour is constantly changing and attacks are evolving at a rapid pace, it is critical to analyse various datasets that have been produced over time using both static and dynamic approaches. This type of research helps in the discovery of the most effective detection algorithm.