Piyush Rai

I am a Computer Science PhD student (2007 - ?) at School of Computing, University of Utah. My advisor is Hal Daumé III. My research interests are into several problems in machine learning and computational statistics. Primarily, I work on designing probabilistic models that can express and reason about uncertainty. Towards this end, I work on problems in Bayesian learning (approximate inference, Bayesian nonparametrics) and probabilistic graphical models.

Besides these, I am also interested in models that can learn with auxiliary information (e.g., auxiliary tasks/domains, unlabeled data, or multiple views), and learning in constrained settings (online/streamed learning).

I did my Bachelors in Computer Science and Engineering from IT-BHU, India. My undergraduate thesis was on skin classification and mean-shift based tracking using motion models.

Recent Papers

Multi-Label Prediction via Sparse Infinite CCA [pdf]
With: Hal Daumé III
NIPS 2009, Vancouver, Canada

Streamed Learning: One-Pass SVMs [pdf]
With: Hal Daumé III, Suresh Venkatasubramanian
IJCAI 2009, Pasadena, CA
(A preliminary version appeared in the Learning Workshop 2008)

The Infinite Hierarchical Factor Regression Model [pdf]
With: Hal Daumé III
NIPS 2008, Vancouver, Canada
(A preliminary version appeared in the Learning Workshop 2008)

Contact:
Piyush Kumar Rai
School of Computing, Room #3190
50 S, Central Campus Drive, University of Utah
Salt Lake City, Utah - 84112-9205

(Office: 4158A, Merrill Engineering Building)