Time-stretching and genetic zooms in OutGene: Detecting Undefined Network Attacks
Jyoti Shekhawat, Ms. Rashmi Khurana · 2022
The study introduces OutGene, a method for detecting malicious behaviour in real time without prior knowledge of assaults or training data. OutGene gathers hosts with comparable behaviour via clustering. We offer the idea of genetic zoom, which involves utilizing a genetic algorithm which select the attributes that is more pertinent to characterize cluster in order to aid human analysts in locating malevolent clusters. Attacks carried out slowly or below thresholds are frequently successful in evading machine learning-based attack detection. We offer the idea of time stretching as a way to identify such stealth attacks. The goal is to discover assaults regardless of how quickly they are carried out by analyzing the stream of events over various time frames. We evaluated OutGene experimentally using data sets currently publicly available and data sets collected at critical military installations. Both genetic zoom and temporal stretching achieved high recall and precision scores, indicating that both are effective.