Research

My research interest is in the broad area of machine learning. The focus of my doctoral work is on learning Mixed Membership Latent Variable Models (sometimes known as topic modeling) that can characterize complex high-dimensional observations including text documents (e.g., Twitter streaming, news articles, etc.), user preferences (pairwise comparisons over products, movies, etc.). Our focus is to develop provable algorithms with computation and sample complexity guarantees under an inevitable separable condition. Please find my dissertation here.

Projects

 

Publications

2015

  • L. Guo, C. Vargo, Z. Pan, W. Ding, and P. Ishwar, “Big Social Data Analytics in Journalism and Mass Communication: Comparing Dictionary-based Text Analysis and Unsupervised Topic Modeling,” Journalism and Mass Communication Quarterly, submitted 2015 (under review).
  • W. Ding, P. Ishwar, and V. Saligrama, “Necessary and Sufficient Conditions and a Provably Efficient Algorithm for Separable Topic Discovery,” in arXiv:1508.05565 [cs.LG] (submitted to IEEE Journal on Special Topics in Signal Processing (JSTSP), Special Issue on Structured Matrices in Signal and Data Processing (under review).)
  • W. Ding, P. Ishwar, and V. Saligrama, “Learning Mixed Membership Mallows Models from Pairwise Comparisons,” in arXiv:1504.00757 [cs. LG]
  • W. Ding, P. Ishwar, and V. Saligrama, “A Topic Modeling Approach to Ranking,”  in Proc. International Conference in Artificial Intelligence and Statistics (AISTATS’15), San Diego, CA, USA, 7-9, May., 2015. [arXiv version]
  • W. Ding, P. Ishwar, and V. Saligrama, “Most Large Topic Models are Approximately Separable,”  in Proc. 10th IEEE International Workshop on Information Theory and Applications (ITA)San Diego, CA, 1-6, Feb., 2015.
  • W. Ding, P. Ishwar, and V. Saligrama, “Learning Shared Rankings From Mixtures of Noisy Pairwise Comparisons,”  in Proc. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP’15), Apr., 2015.

2014

2013