SDS Colloquium, Speaker Travis Wheeler

Title: Representation Learning for improved understanding of proteins

Abstract: Deep Learning techniques have produced remarkable breakthroughs across a diverse
space of challenging problems, including representation and understanding of protein
structures and their interactions with small molecule drugs. In my talk, I will describe
new lightweight Deep Learning models that improve accuracy (over state-of-art
foundation models) on tasks involving sequence relationship detection and predicting
protein-drug interaction, while at the same time yielding orders of magnitude speed
gains. I will also discuss our efforts to gather massive new data sets to train the next
generation of Deep Learning methods for virtual drug screening.