Zero-Shot Learning and Learning with Limited Data
Deep neural networks and fully supervised training has had tremendous success over the last decade, and in no small part due to the large-scale availability of annotated ground-truthed data. Nevertheless, in many applications where new, unanticipated and heretofore unseen scenarios arise, we are unlikely to find sufficient data due to acquisition costs or real-time considerations, severely limiting the scope of existing supervised learning methods. In this talk we will motivate the problems we face in this context by drawing upon examples from our work on zero-shot recognition, crowd-sourcing, and feedback-limited sequential learning.
Venkatesh Saligrama is a faculty member in the Department of Electrical and Computer Engineering and Department of Computer Science (by courtesy) at Boston University. He holds a PhD from MIT. His research interests are in Machine Learning and its applications. He has edited a book on Networked Sensing, Information and Control. He has served as an Associate Editor for IEEE Transactions on Information Theory, IEEE Transactions on Signal Processing and edited special issues and books on Networked Sensing, Detection and Estimation. He is currently serving as the Vice-Chair of Big Data Special Interest Group for IEEE SPS society. He is an IEEE Fellow and recipient of several awards including the Presidential Early Career Award (PECASE), ONR Young Investigator Award, the NSF Career Award and a NIPS 2014 workshop best student paper award on Analysis of Ranking Data. More information about his work is available at http://sites.bu.edu/da.
Host: Professor Donald Brown