I am a Postdoctoral Associate at the MIT Sloan School of
Management sponsored by Prof. Sinan Aral.
I am a data scientist and am interested in understanding the impact of digital technologies on our lives, and economy. My research broadly involves making careful empirical measurements (both causal and predictive) from big data, and draws on ideas from Computer Science, Statistics, Social Sciences, and Applied Econometrics.
My current research is supported by collaborations with The Boston Globe, The New York Times, and NerdWallet.
I hold an A.M. in Statistics and M.S.E & Ph.D. in Computer &
Information Science (CIS), all from the University of
Pennsylvania, where I was advised by Profs. Lyle Ungar and Dean Foster.
My doctoral dissertation won the Morris & Dorothy Rubinoff Best Dissertation Award. My doctoral work, which was on statistical Machine Learning & Natural Language Processing (NLP), proposed simple linear methods that are both statistically and computationally efficient and give accuracies comparable to or better than state-of-the-art neural networks based methods on text data.
Other parts of my Ph.D research provided theoretical and empirical contributions to statistical modeling such as:
- Showing that Principal Component Regression (PCR) can never be much worse than Ridge
Regression, but can be infinitely better. (Citation: JMLR 2013)
- Extending Spectral Learning algorithms to trees, particularly the ones arising from
dependency structures in natural language. (Citation: EMNLP 2012)
- Formulating optimal information theoretic coding schemes for feature/covariate
selection. (Citations: JMLR 2011, ICDM 2008, ECML 2009, ACL 2009)
- Information Extraction from structured Web data in the
presence of a). Domain-specific constraints on the records to be
extracted and b). Scarce availability of labeled training data. (Citations: CIKM 2011, AISTATS 2012)
For more details you might want to have a look at my Publication List (Google Scholar Profile) or CV.
77 Massachusetts Avenue
Cambridge, MA, U.S.A