University of Guelph
I work on deep learning theory and methodology, so that models can be deployed with useful performance guarantees in practice. I’m particularly interested in the application of information theory to deep learning, and in the service of Engineering standards for its use.
My current research interests are motivated by the “adversarial examples” phenomenon, which limits opportunities for the responsible deployment of deep learning models in performance-critical settings, e.g., commercial autopilots, or medical imaging. More generally, I am interested in privacy and security, causal discovery, and embedded systems.
|Nov 14, 2019||I will lead breakout sessions on robustness to distribution shift and information-theoretic approaches in Deep Learning at the upcoming Pan-Canadian Self-Organizing Conference on Machine Learning.|
|Nov 14, 2019||I will present work on batch normalization at the Toronto Machine Learning Summit.|
|Sep 3, 2019||Top 400 reviewer award for NeurIPS 2019.|
|May 22, 2019||New work Batch Normalization is a Cause of Adversarial Vulnerability to appear in the Workshop on Identifying and Understanding Deep Learning Phenomena at ICML 2019.|
|Mar 25, 2019||Awarded a Dean’s Graduate Scholarship from the College of Engineering & Physical Sciences at the University of Guelph, and an Ontario Graduate Scholarship.|