P. Balamurugan

Update:

About My Research..

I am a PhD student at CSA deparment, IISc, advised by Dr. Shirish K. Shevade. My thesis focuses on developing efficient algorithms for structured prediction and sparse classification problems. I have been involved in developing optimization methods for structural SVMs, CRFs, SVMs and logistic regression.

Software

The code for an efficient sequential dual optimization method for structural SVMs is made available for free use. Please find the code at this link.

Reference paper:

A Sequential Dual Method for Structural SVMs.

Publications

• 2014

1. Scalable sequential alternating proximal methods for sparse structural SVMs and CRFs.

   P.Balamurugan, Shirish Shevade, T.Ravindra Babu.
   In Knowledge and Information Systems. Vol. 38(3), pages: 599-621. March, 2014.
   

• 2013

1. Large-Scale Elastic Net Regularized Linear Classification SVMs and Logistic Regression.

   P. Balamurugan.
   In ICDM, 2013. Acceptance Rate: 19.65% (159/809)
   
   

2. Optimizing F-Measure With Non-Convex Loss and Sparse Linear Classifiers.

   Punya Murthy Chinta, P. Balamurugan, Shirish Shevade, M. Narasimha Murty.
   In IJCNN, 2013.
   

• 2012

1. Sequential Alternating Proximal Method for Scalable Sparse Structural SVMs.

   P.Balamurugan, Shirish Shevade, T.Ravindra Babu.
   (Full Paper) In IEEE International Conference on Data Mining (ICDM), 2012. Acceptance Rate for full papers: 10.71% (81/756).
   (One of the Best Papers invited for KAIS journal). [pdf]
   
   

2. Efficient Algorithms for Linear Summed Error Structural SVMs.

   P.Balamurugan, Shirish Shevade, T.Ravindra Babu.
   In International Joint Conference on Neural Networks (IJCNN), 2012. [pdf]
   

• 2011

1. A Sequential Dual Method for Structural SVMs.

   P.Balamurugan, Shirish Shevade, S.Sundararajan, S.S.Keerthi.
   In SIAM International Conference on Data Mining (SDM), 2011. [pdf]

Workshop Papers

• Efficient Variational Inference for Gaussian Process Structured Prediction.

  P. K. Srijith, P.Balamurugan, Shirish Shevade.
  In NIPS workshop on Advances in Variational Inference, 2014.
  

Papers under review

• A Simple Label Switching Algorithm for Semi-supervised Structural SVMs.

  P.Balamurugan, Shirish Shevade, S.Sundararajan.
  Submitted.
  

Technical Report (In Arxiv)

• An Empirical Evaluation of Sequence-tagging Trainers.

  P.Balamurugan, Shirish Shevade, S.Sundararajan, S.S.Keerthi, 2013. [arXiv link]

Education

• PhD Student at Department of CSA, IISc, Bangalore. (since January 2010)

• M.E.(Computer Science Engineering), Department of CSA, IISc, Bangalore. (2007-2009)

• B.Tech (Information Technology), Department of IT, Thiagarajar College of Engineering, Madurai. (2000-2004)

Teaching Assistance

• Teaching Assistant for Probability and Statistics course by Dr.Indrajit Bhattacharya. (August-December 2011)

Work Experience

• Research Intern at IBM India Research Lab, Bangalore, during May-August, 2013. Mentor: Dr. Vikas C. Raykar.

• Project Assistant under Dr. Shirish Shevade during August 2009 - December 2009. Worked on "Efficient Algorithms for Structured Prediction".

• Assistant Systems Engineer at Tata Consultancy Services during 2004 - 2007.

M.E. Thesis

• L1-Norm Structural Classification SVMs.

Other Activities

• An organizer for CSA Undergraduate Summer School, 2013.

• Designed a movie-rating recommender system for CSA Open Days, 2013. (Joint work with Anusha Posinasetty, with inputs from Srijith P.K. and Divya Padmanabhan)

• Gave a talk on "Basics of Classification and Support Vector Machines" during CSA Undergraduate Summer School, 2012.