Mudassir Shabbir
Research Assistant Professor of Computer Science
Research Focus
Graph
Theoretic Methods in Cyber Physical Systems, Graph Machine Learning, Robustness
and Resilience in CPS
Biography
The focus of Dr. Shabbir's research work is on the broad area of the application of Graph Theory in networks problems in data science and control systems and an AI-based design space exploration in cyber-physical systems. As part of the machine learning application of graph theory, he works on the design of robust graph representations based on graph contractability, Kirchhoff index, and network controllability. The representations are designed with a focus on objectives like permutation-invariance, multi-scale information encapsulation, state-of-the-art classification accuracies, and model interpretability. He also works on the resilience of distributing learning methods in adversarial environments.
In control networks, he is interested in exploring various desirable properties of a network like how to control the states of individual nodes with very few control nodes, how a network behaves in presence of a few adversarial nodes, and how small changes in the network topology affect control behavior of a system. He also studies the effects of graph topology on achieving these properties, the trade-offs between them, and their computational aspect
Previously, he has developed new methods for the characterization and computation of succinct representations of large data sets with applications in nonparametric statistical analysis. In the past, Mudassir has been associated with Los Alamos National Labs (LANL), NM, Bloomberg L.P. New York, NY, and Rutgers University, NJ.