In the era of information systems and internet there is more concern rising towards information security in day-to-day life, along with the availability of the vulnerability assessment mechanisms to identifying the electronic attacks. Anomaly detection is the process of attempting to identify instances of attacks by comparing current activity against the expected actions of intruder. Machine learning based intrusion detection has the ability to change its execution plan as it obtains new information. This paper is to provide a comprehensive review of some machine learning based techniques have been applied to AIDS (with identifying their main pros and cons.