Abstract

The rapid advancements in the communication technology and information exchange in cyber space has led to the issue of cyber-attacks. As the attackers are finding new techniques of designing the cyber-attacks, there is an urgent need to design a robust cyber-attack detection and mitigation system. This study explores the various Artificial Intelligence (AI) and Machine Learning (ML) based approaches for the detection of cyber-attacks. The different threats and risks have been categorized into three main types: network-based, host-based, and application-level attacks. 

Various AI/ML algorithms such as Random Forest(RF), Support Vector Machines (SVM), Convolutional Neural Networks (CNN), and LSTM-based architectures used on existing datasets are compared based on their detection capabilities, accuracy, and application contexts.

The paper also tries to identify challenges related to the quality of datasets, model interpretability, and the detection of zero-day attacks in an attempt to highlight the need for AI-driven smart, hybrid and adaptive solutions. The survey conducted in this paper serves as a foundation for researchers and practitioners who are aiming to develop robust intrusion detection systems to mitigate advance cyber-attacks.

Statistics