Krishnakumar Balasubramanian


gt College of Computing (CSE)
Georgia Institute of Technology
Email:
Office: 1305 KACB
CV: [PDF]



I am currently a PhD student at College of Computing, Georgia Tech. Prior to this, I obtained my Masters degree in 2010. I work with Prof. Guy Lebanon. I received my Bachelors degree from Madras Institute of Technology in 2008. As an undergrad, I worked on wireless communications and information theory. Yes, I select text while reading :)


Research

I am interested in all aspects of mathematical statistics (classical and modern) and statistical machine learning. Current research interests include developing and analyzing:
  • estimation procedures for high dimensional data
  • labeling complexity reduction methods (for classification, regression and structured prediction)
Thanks to Facebook, I will be supported by Facebook Ph.D. Fellowship for 2013-14.

Papers
    Journals

  • Unsupervised Supervised Learning II: Margin-Based Classification without Labels. [Link]
    K. Balasubramanian, P. Donmez, G. Lebanon.
    Journal of Machine Learning Research (JMLR) 12, pp. 3119-3145, 2011.

  • Unsupervised Supervised Learning I: Estimating Classification and Regression Errors without Labels. [Link]
    P. Donmez, G. Lebanon, K. Balasubramanian.
    Journal of Machine Learning Research (JMLR) 11, pp. 1323-1351, 2010.

  • Conferences

  • High-dimensional Joint Sparsity Random Effects Model for Multi-task Learning. [PDF]
    K. Balasubramanian, K. Yu, T. Zhang.
    Proc. of the 29th Conference on Uncertainty in Artificial Intelligence (UAI), 2013 (to appear).

  • Ultrahigh Dimensional Feature Screening via RKHS Embeddings. [PDF]
    K. Balasubramanian, B. K. Sriperumbudur, G. Lebanon.
    Proc. of the 16th International Conference on Artificial Intelligence and Statistics (AISTATS), 2013 (to appear).

  • Smooth Sparse Coding via Marginal Regression for Learning Sparse Representations. [PDF]
    K. Balasubramanian, K. Yu, G. Lebanon.
    Proc. of the 30th International Conference on Machine Learning (ICML), 2013 (to appear).

  • A Landmark Selection Approach for Multi-output Prediction Problems. [PDF]
    K. Balasubramanian, G. Lebanon.
    Proc. of the 29th International Conference on Machine Learning (ICML), 2012.

  • Unsupervised Supervised Learning II: Margin-Based Classification without Labels. [PDF] [arXiv link]
    K. Balasubramanian, P. Donmez, G. Lebanon.
    Proc. of the 14th International Conference on Artificial Intelligence and Statistics (AISTATS), 2011.

  • Asymptotic Analysis of Generative Semi-Supervised Learning. [PDF] [arXiv link]
    J. V. Dillon, K. Balasubramanian, G. Lebanon.
    Proc. of the 27th International Conference on Machine Learning (ICML), 2010.

  • Dimensionality Reduction for Text using Domain Knowledge. [PDF] [arXiv link]
    Y. Mao, K. Balasubramanian, G. Lebanon.
    Proc. of the 23rd International Conference on Computational Linguistics (COLING), 2010.

  • A Fast Algorithm for Nonnegative Tensor Factorization using Block Coordinate Descent
    and an Active-set-type Method,
    . [PDF]
    K. Balasubramanian, J. Kim, A. Puretskiy, M. W. Berry, H. Park.
    Proc. of the 2010 Text Mining Workshop: SIAM International Conference on Data Mining, 2010.

  • Thesis

  • Learning without labels and nonnegative tensor factorization. [Link]
    K. Balasubramanian.
    Master's Thesis, Georgia Institute of Technology, 2010.


Selected Courses

TA Experience