Lecture: Multilevel Feature Integration for Semantic Labeling of Images, Greg Shakhnarovich, March 10

Presented by the Research Computing Center

Multilevel Feature Integration for Semantic Labeling of Images

Greg Shakhnarovich, Assistant Professor, Toyota Technical Institute at Chicago

Tuesday, March 10, 2015 | 3-4:30 p.m. Crerar Library, Kathleen A. Zar Room

Greg Shakhnarovich will speak about the problem of semantic labeling of images, one of the core problems in computer vision. Recent advances in computational learning methods, in particular deep neural networks, have led to a radical shift in this and other areas of visual computing. He will discuss the rapidly evolving state of the art for semantic labeling, and present his recent and ongoing work on leveraging the power of deep networks in a model that integrates information across levels of increasing spatial extent and complexity (the “zoom-out” framework). His talk will also address the computational challenges that arise in the context of training such models.

Biography: Since February 2008, Greg Shakhnarovich has been an Assistant Professor at TTI-Chicago, a philanthropically endowed academic computer science institute located on the University of Chicago campus. He also hold a part-time faculty appointment at the University of Chicago Department of Computer Science. Prior to coming to TTI-Chicago, Greg was a post-doctoral researcher at the Department of Computer Science of Brown University where he worked with Michael Black. He received his PhD at MIT where he worked at CSAIL with Trevor Darrell on computer vision and machine learning. Greg’s thesis topic was Learning Task-Specific Similarity.

For more information contact RCC at info@rcc.uchicago.edu or 773-795-2667