With the increasing prevalence of major cyber incidents worldwide, safeguarding user data has become more critical than ever in today’s digital era. Cyberattacks can have devastating impacts across various sectors, underscoring the need for robust security measures. Rapid growth and complexity of network systems and internet have led to the development of numerous techniques for detecting intrusions. So, machine learning along with meta-heuristic techniques offers a promising solution for developing a reliable and efficient IDS. This paper discusses about different classifiers like single, hybrid, and ensemble encompassing various research articles in this domain. It also provides a comprehensive review of notable recent works, advantages of hybrid and ensemble classifiers over single classifier, updated datasets, and discusses performance metrics utilized in the development of IDS. However, in order to deal with research challenges like imbalanced datasets, high computational cost, etc., a method is proposed to detect different attacks on high volumes of data over the network.