RSTCON 2025 - Attributions For ML-Based ICS Anomaly Detection - Clement Fung

RSTCON is an annual technical security conference aimed at resetting our focus to cutting-edge research, exploitation, and tradecraft targeting the sensors, systems, and architectures utilized by critical industry.

In this presentation from 2025, Clement Fung explores the critical challenges of detecting and diagnosing attacks on Industrial Control Systems (ICS) that power essential infrastructure (e.g., power plants and water treatment facilities) using state-of-the-art attribution methods for machine-learning-based anomaly detectors, revealing why they often fall short in real-world scenarios and offering practical recommendations, including the use of attribution ensembles for improved results.

Clement Fung is a postdoctoral researcher at Carnegie Mellon University’s Software and Societal Systems Department and CyLab, where he recently completed his Ph.D. (Fall 2025) in Societal Computing. Fung's research focuses on the intersection of security, machine learning, and cyber-physical systems, with a particular emphasis on improving anomaly detection for securing industrial control systems.

See https://rstcon.org/ for more information.

Видео RSTCON 2025 - Attributions For ML-Based ICS Anomaly Detection - Clement Fung канала RSTCON