Abstract: 

I will present new computer vision algorithms to learn complex morphologies and phenotypes that are important for human diseases. I will illustrate this approach with examples that capture physical scales from macro to micro: 1) video-based AI to assess heart function (Ouyang et al Nature 2020), 2) generating spatial transcriptomics from histology images (He et al Nature BME 2020), and 3) making genome editing safer (Leenay et al Nature Biotech 2019). Throughout the talk I'll illustrate the general principles and tools for human-compatible ML that we’ve developed to enable these technologies (Ghorbani et al. ICML 2020, Abid et al. Nature MI 2020). These references are available at www.james-zou.com.

Bio: 

James Zou is an assistant professor of biomedical data science and, by courtesy, of CS and EE at Stanford University. Professor Zou develops novel machine and deep learning algorithms that have strong statistical guarantees; several of his methods are currently being used by biotech companies. He also works on questions important for the broader impacts of AI—e.g. interpretations, robustness, transparency—and on biotech and health applications. He has received several best paper awards, a Google Faculty Award, a Tencent AI award, and is a Chan-Zuckerberg Investigator. He is also the faculty director of Stanford AI for Health program and is a member of the Stanford AI Lab.