Intrusion detection in forensics based on machine learning techniques: a review. (8th November 2021)

• Record Type: Journal Article
• Authors: Bistouni, Fathollah; Jahanshahi, Mohsen; Tee, Kong Fah
• Is Part Of: International journal of forensic engineering. Volume 5:Number 2(2021)
• Page Start: 126
• Page End: 156
• Publication Date: 2021-11-08
• Subjects: intrusion detection -- machine learning -- forensics -- data mining -- supervised learning -- unsupervised -- semi-supervised; Forensic engineering -- Periodicals; 624.176
• Languages: English
• ISSNs: 1744-9944

Abstract

Penetration into various systems, including information, organisations, banks and other systems has become a challenge. Intrusion detection systems (IDS) today have a great impact on detecting attacks and intrusions on many systems including forensics, and a nuclear design that can accurately perform the intrusion detection process is crucial. This paper discusses machine learning techniques of IDS design and implementation in forensics. In general, machine learning is categorised into three general categories: supervised, unsupervised and semi-supervised learning to detect intrusion. In each of these categories, techniques have been put forward that each one with its outstanding capabilities and features can be effective in detecting intrusion. Surveys and analyses show that supervised techniques have higher accuracy and capability to detect intrusions into the IDS.