Department of Mechanical Engineering Iowa State University
Spatiotemporal Graphical Modeling for Complex Cyber-Physical Systems
Modern distributed cyber-physical systems (CPSs) typically encounter a large variety of physical faults and cyber anomalies and in many cases, they are vulnerable to catastrophic fault propagation scenarios due to strong connectivity among the sub-systems. Majority of the state-of-the art techniques for detecting such anomalies depend on the knowledge of fault/attack characteristics. However, it is quite infeasible to have a comprehensive knowledge of faults/anomalies in real-life large CPSs. This talk will discuss a new datadriven modeling framework for system-wide anomaly detection in an unsupervised manner. The framework is based on a spatiotemporal feature extraction scheme built on the concept of symbolic dynamics for discovering and representing causal interactions among the subsystems of a CPS. The extracted spatiotemporal features are then used to learn system-wide patterns via a Restricted Boltzmann Machine (RBM) which is an energy based probabilistic graphical model. The anomaly detection and root-cause analysis process developed here aims to detect and isolate low probability events by using the concept of free energy of RBM.
Refreshments at 3:45 on in Snedecor 2101.