Research Projects

The Internet's growing importance in human life has led to a global issue of cyber attacks. Intrusion detection systems have been developed to mitigate these threats, but they are not always accurate due to unknown attacks and the use of rule-based or anomaly-based monitoring devices. Machine learning-based systems have potential solutions, but their limitations are based on the training classifier. The proposed research aims to provide global-scale results in areas such as using and comparing databases of CA, working with number CA heterogeneity, using machine learning classifiers, and merging results of various classifiers. The project aims to increase intrusion detection rates for rare and high evasion rate intrusions by creating hybrid platform detectors and classifiers based on machine learning and neural networks. The major outcomes include a system design capable of real-time detection of intrusion traces, a robust algorithm efficient in time and precision space, and joint publication. The project will target Indian and Slovenian researchers and students in the first year, developing intelligent methods for infrequent and high evasion rate intrusion detection and classification, and developing a web-portal for methods testing.