AI-ENABLED CYBER INTELLIGENCE AND SECURITY RESEARCH
The CIS department has engaged in research to harness machine learning (ML) and artificial intelligence (AI) for enhancing threat intelligence and cybersecurity.
The establishment of advanced facilities in the department’s Cyber Lab and Cyber Gym, such as a deep learning server, is expected to accelerate future research in this domain. AI- and ML-Enabled Cyber Aids for Crew This project aims to use ML and AI to enhance the situational awareness and decision making of pilots and air traffic controllers facing cyber threats. The focus is on integrating crew-based cybersecurity engagement as a crucial layer to bolster the resilience of aircraft and ATC systems against cyber and cyber-physical attacks
leading to drone misbehavior, and the limitations of traditional security solutions, such as virtual private networks. The project explores the use of ML for intrusion detection. ML Applications in Threat Intelligence This project explores ML techniques to identify and classify potential malware files. The project compares the assembly (machine code) of files, eliminating the need to execute malicious code. In the realm of AI-assisted intrusion detection, large data models can help enhance the effectiveness of industry-standard tools. The incorporation of ML aims to mitigate the risks of false results.
AI- and ML-Enabled Penetration Testing
Penetration testing is an effective method used to proactively address network vulnerabilities. However, existing challenges include time-consuming processes, susceptibility to human error and a shortage of qualified professionals. To overcome these challenges, this project proposes integrating ML techniques to automate pentest processes and interpret results. The approach is a collaboration between ML and human testers to enhance quality in shorter timeframes. ML-Based Intrusion Detection for Drones and Communications This project focuses on cybersecurity challenges faced by uncrewed aircraft systems, specifically focusing on the vulnerability of their command-and-control network protocols. It highlights the potential risks, such as packet injection
19
Made with FlippingBook - PDF hosting