Unknown Threat Detection using AI, ML and DL methods

Cyber threats in any organization is a constant concern for the IT teams. Many tools available in the market only detect known threats and many uses fixed signatures which hackers and attackers are unlikely to reuse. Any rule-based engine deployed is ineffective given the nature of manifestation of threat attacks hackers use these days. If a system can learn the network over time from what is expected and what is not expected it can provide better detection of unusual traffic patterns in the network which are threat oriented. This research explores a new way of detecting unknown threats which are happening for the first time in a IT network by using machine learning or behavior analysis of historical data and also certain probabilistic functions to qualify a cyber threat. It depends of historical behavior learning from an IT network where behavior patterns are learnt over time and used to detect anomalous conditions and qualifying them further as potential threats by different means