Abstract: Many domains in the real world are richly structured, containing a diverse set of objects, related to each other in a variety of ways. For example, a living cell contains a rich network of interacting genes, that come together to perform key functions. A robot scan of a physical environment contains classes of objects such as people, vehicles, trees, or buildings, each of which might itself be a structured object. However, most applications of machine learning aim to simplify the problem by considering objects in the domain as independent instances from a single distribution. In this talk, I aim to show that one can gain from modeling both the dependencies arising from the relationships between objects, and the rich structure of the similarities and differences between them. The first part of the talk will describe a rich language, based on probabilistic graphical models, which allows us to model the rich network of dependencies between related objects; we show how to learn such models from data and how to use the learned model both for knowledge discovery and for reasoning about new instances. The second part of the talk focuses on methods for learning the similarities and differences between related yet diverse classes of objects (such as different types of animals), so as to allow information learned for one class to transfer to another. I will describe applications of this framework to two main tasks: modeling objects in the physical world, and recognizing them in laser range scans and in images; and inferring a network of regulatory interactions in a cell, and how this network is perturbed by individual genotype.

Short Bio: Daphne Koller received her BSc and MSc degrees from the Hebrew University of Jerusalem, Israel, and her PhD from Stanford University in 1993. After a two-year postdoc at Berkeley, she returned to Stanford, where she is now an Associate Professor in the Computer Science Department. Her main research interest is in creating large-scale systems that reason and act under uncertainty, using techniques from probability theory, decision theory and economics. Daphne Koller is the author of over 100 refereed publications, which have appeared in venues spanning Science, Nature Genetics, the Journal of Games and Economic Behavior, and a variety of conferences and journals in AI and Computer Science. She was the co-chair of the UAI 2001 conference, and has served on numerous program committees and as associate editor of the Journal of Artificial Intelligence Research and of the Machine Learning Journal. She was awarded the Arthur Samuel Thesis Award in 1994, the Sloan Foundation Faculty Fellowship in 1996, the ONR Young Investigator Award in 1998, the Presidential Early Career Award for Scientists and Engineers (PECASE) in 1999, the IJCAI Computers and Thought Award in 2001, the Cox Medal for excellence in fostering undergraduate research at Stanford in 2003, and the MacArthur Foundation Fellowship in 2004.