Turning Big Data into Core Data

Jeffrey Bilmes

University of Washington

Abstract

The wealth of available data has been a problem for human consumers of information for decades.  At this point, the explosion of available data is a problem even for computer consumers of information. In this talk, we will discuss how submodular function optimization can address such issues. For example, the generic problem of joint active/semi-supervised learning, and in particular the problems of document summarization and data subset selection in speech recognition, can be successfully addressed using submodular optimization. We will also survey a number of recent problems in machine learning where submodular functions are appropriate.  The talk will include sufficient background to make it accessible to everyone.

Biography  

Prof. Jeff Bilmes is a professor at the Department of Electrical Engineering at the University of Washington, Seattle Washington, and also an adjunct professor in Computer Science & Engineering and the department of Linguistics. He received his Ph.D. from the Computer Science Division of the department of Electrical Engineering and Computer Science, University of California in Berkeley, and was also a research scientist at the International Computer Science Institute, in Berkeley.  His primary research interests lie in statistical modeling, dynamic graphical models, information theory, submodularity in combinatorial optimization, and high performance computing, all with applications in machine learning. Additional applications he has worked on include speech recognition, natural language processing, computer vision, bioinformatics, human/computer interfaces, and audio/music processing.