[cs-talks] IVC seminar - Thursday April 14
cs at bu.edu
Tue Apr 12 10:51:20 EDT 2016
Algorithms and new applications for determinantal point processes
April 14, 2-3pm
Many real-world inference problems are, at their core, subset selection problems. Probabilistic models for such scenarios rely on having sufficiently accurate yet tractable distributions over discrete sets. We focus on sub-families of such distributions whose special mathematical properties are the basis for fast algorithms. As a specific example, Determinantal Point Processes (DPPs) have recently become popular in machine learning, as elegant and tractable probabilistic models of diversity.
I will outline known and new applications of DPPs for machine learning, from diverse subset selection to large-scale learning and variational inference over combinatorial objects. These problems call for fast algorithms. While sampling from DPPs is possible in polynomial time, the associated algorithm is not practical for large data. In the second part of the talk, I will outline ideas for faster sampling that build on new insights for algorithms that compute bilinear inverse forms. These results have applications beyond DPPs, including sensing with Gaussian Processes and submodular maximization.
This is joint work with Chengtao Li, Suvrit Sra, Josip Djolonga and Andreas Krause.
Stefanie Jegelka is an X-Consortium Career Development Assistant Professor in the Department of EECS at MIT, where she is a member of CSAIL and IDSS. Before joining MIT, she was a postdoctoral researcher at UC Berkeley. She obtained a Ph.D. from ETH Zurich in collaboration with the Max Planck Institutes in Tuebingen, Germany. She has received an NSF CAREER Award, a Google research award, the German Pattern Recognition Award and an ICML Best Paper Award. Her research interests lie in algorithmic machine learning and various application areas.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the cs-talks