Wednesday, October 13, 2021 - 2:00pm to 3:00pm
Event Calendar Category
Uncategorized
Speaker Name
Matus Telgarsky
Affiliation
University of Illinois, Urbana-Champaign
Zoom meeting id
95607814981
Join Zoom meeting
https://mit.zoom.us/s/95607814981
Abstract
The talk starts by surveying the implicit bias of standard descent methods, with some emphasis on proof schemes. The remainder of the talk will focus on an implicit bias, specifically an alignment phenomenon, in deep learning and in actor-critic algorithms.
Joint work with Yuzheng Hu and Ziwei Ji.
Biography
Matus Telgarsky is an assistant professor at the University of Illinois, Urbana-Champaign, specializing in deep learning theory. He was fortunate to receive a PhD at UCSD under Sanjoy Dasgupta. Other highlights include: co-founding, in 2017, the Midwest ML Symposium (MMLS) with Po-Ling Loh; receiving a 2018 NSF CAREER award; co-organizing a Simons Insititute program on deep learning theory.