Colloquium - Dave Zhao, "Representation Learning in Computational Genomics"

Abstract: 
There is tremendous untapped opportunity for innovation at the intersection of computer science and genomic biology. For example, a major class of questions in genomics aim to identify and understand the latent biological factors underlying observed high-throughput molecular readouts. Representation learning techniques from computer science offer a powerful solution. I will describe three methods we have developed, in response to new problems arising from recently developed spatial genomics technologies, that leverage manifold learning, network embedding, sparse matrix decomposition, and causal representation learning. I will then highlight additional areas ripe for new ideas. 

Bio:
Dave Zhao is an Associate Professor of Statistics at the University of Illinois Urbana-Champaign. His research focuses on nonparametric empirical Bayes methods, high-dimensional statistics, and statistical genomics. He is also the Director of Computational Genomics at the Carl R. Woese Institute for Genomic Biology, where he leads initiatives to identify, initiate, and develop cutting-edge collaborations between computational and biological researchers.