Research

My general research focus is on optimization and graphical models. I enjoy research that focuses on application areas, but has a strong theoretical component. My work in this vein has thus far involved Learning to Rank and Human Computation as application areas. Please see my publications for details on these. I am also interested in and have done courses/projects involving Large Scale Learning (both high dimensional and big data settings), Time Series Analysis and NLP.

My current and upcoming work applies functional analysis and infinite dimensional optimization ideas, eg. our recent paper on A Representation Theory for Ranking Functions (see publications below). We are currently appyling these ideas to graphical models, contextual bandits and other problems.

Publications

  • A Representation Theory for Ranking Functions [pdf][supp] [poster]
    H. Pareek, P. Ravikumar.
    In Advances in Neural Information Processing Systems (NIPS) 27, 2014.
  • Human Boosting [supp] [slides] [poster]
    H. Pareek, P. Ravikumar.
    In International Conference on Machine Learning (ICML) 30, 2013.
  • Distributional Rank Aggregation, and an Axiomatic Analysis [pdf] [Supp] [slides] [poster]
    A. Prasad*, H. Pareek*, P. Ravikumar
    In International Conference on Machine Learning (ICML) 32, 2015
  • Tracking with ranked signals [supp] [slides] [poster]
    Tianyang Li, Harsh Pareek, Pradeep Ravikumar, Dhruv Balwada, Kevin Speer.
    UAI 2015. (selected for plenary presentation)