Starting September 2013, I am a postdoctoral researcher at Duke University. Prior to that, I spent a year as a postdoc with the Division of Statistics and Computation at UT Austin. I completed my PhD in Computer Science in August 2012 from the School of Computing, University of Utah. My advisor was Hal Daumé III.

My research interests are into several problems in machine learning and computational statistics. My PhD dissertation was about designing nonparametric Bayesian models for learning low-dimensional latent structures from high-dimensional data (e.g., for learning latent features), and from multiple related learning tasks (e.g., for learning latent shared predictive structures). In addition, I also work on designing efficient inference methods for nonparametric Bayesian models, and probabilistic graphical models (structure learning and approximate inference).

Apart from these, some of my other previous works have been on designing label-efficient algorithms (e.g., domain adaptation, transfer learning, semi-supervised learning, active learning), learning with heterogeneous/multi-modal data (multiview learning), and large-scale learning (online learning, feature hashing). 

Teaching:
- Machine Learning (Fall 2011) 
- Machine Learning Seminar (Fall 2010) 

Publications

Conferences/Journals

Refereed Workshop and Misc. Papers

  • Enhancing the Perception of Collaborative Actions with Virtual Gestures [pdf][bib]
    With: Patrick Horain, José Marques Soares, and André Bideau
    AFVR 2006, Rocquencourt, France

  • Virtually Enhancing the Perception of User Actions [pdf][bib]
    With: Patrick Horain, José Marques Soares, and André Bideau
    ICAT 2005, Christchurch, New Zealand