Natural language research announcements

["Samdani, Rajhans" <rsamdan2 AT illinois.edu>]

This talk should be of great interest to people working in Machine Learning.

---

From: cs-grads-bounces AT cs.uiuc.edu [cs-grads-bounces AT cs.uiuc.edu] on behalf of Gustafson, Julie [jdg5 AT illinois.edu]
Sent: Monday, March 04, 2013 2:08 PM
To: Announce; Clerical; Faculty; Grads; Postdocs
Cc: Cimarusti, Scott; McElroy, Rhonda Kay
Subject: [cs-grads] CS Special Seminar -Daniel Hsu - March 7 @ 11 am

 

 

University of Illinois at Urbana-Champaign

Department of Computer Science

Thomas M. Siebel Center for Computer Science

201 North Goodwin Avenue

Urbana, Illinois 61801-2302 USA

 

Computer Science

Special Seminar

 

Fast learning algorithms for discovering the hidden structure in data

 

 

 

 

Guest Speaker: Daniel Hsu, Microsoft Research New England

Date/time: Thursday, March 7 @ 11 am

Location: 2405 Siebel Center

 

Abstract:

A major challenge in machine learning is to reliably and automatically discover hidden structure in data with minimal human intervention.  For instance, one may be interested in understanding the stratification of a population into subgroups, the thematic make-up of a collection of documents, or the dynamical process governing a complex time series.  Many of the core statistical estimation problems for these applications are, in general, provably intractable for both computational and statistical reasons; and therefore progress is made by shifting the focus to realistic instances that rule out the intractable cases.  In this talk, I'll describe a general computational approach for correctly estimating a wide class of statistical models, including Gaussian mixture models, Hidden Markov models, Latent Dirichlet Allocation, Probabilistic Context Free Grammars, and several more.  The key idea is to exploit the structure of low-order correlations that is present in high-dimensional data.  The scope of the new approach extends beyond the purview of previous algorithms; and it leads to both new theoretical guarantees for unsupervised machine learning, as well as fast and practical algorithms for large-scale data analysis.

 

 

Bio:

Daniel Hsu is a postdoc at Microsoft Research New England.  Previously, he was a postdoc with the Department of Statistics at Rutgers University and the Department of Statistics at the University of Pennsylvania from 2010 to 2011, supervised by Tong Zhang and Sham M. Kakade.  He received his Ph.D. in Computer Science in 2010 from UC San Diego, where he was advised by Sanjoy Dasgupta; and his B.S. in Computer Science and Engineering in 2004 from UC Berkeley.  His research interests are in algorithmic statistics and machine learning.