Sahand Negahban

sahandn ||at|| mit {{dot}} edu
EECS Department
MIT
77 Massachusetts Avenue
Cambridge, MA 02139

About me

I am currently a postdoc working with Professor Devavrat Shah at MIT. Please see my CV for more details.

Prior to that, I received my PhD in Electrical Engineering and Computer Sciences with a designated emphasis in Communication, Computation, and Statistics from the University of California, Berkeley working with Professor Martin Wainwright. In December 2011 I received an M.A. in Statistics from UC Berkeley and, in May 2006, I received my B.S. in Electrical Engineering and Computer Sciences from Cal.

In May, 2011 I received the Yahoo! KSC award and gratefully acknowledge the support.

Research Focus

The focus of my research is to develop theoretically sound methods, which are both com- putationally and statistically efficient, for extracting information from large datasets. A salient feature of my work has been to understand how hidden low-complexity struc- ture in large datasets can be used to develop computationally and statistically efficient methods for extracting meaningful information for high-dimensional estimation prob- lems. My work borrows from and improves upon tools of statistical signal processing, machine learning, probability and convex optimization.

Publications

• S. Negahban, P. Ravikumar, M. J. Wainwright and B. Yu. (2012) A unified framework for the analysis of regularized $M$-estimators. Statistical Science, 27(4): 538--557, December, 2012.

  Preliminary Versions

  Technical Report Number 797.

  Advances in Neural Information Processing Systems, December, 2009 [PDF] Oral Presentation given at NIPS: [Slides]

• S. Negahban, S. Oh, and D. Shah. Iterative Ranking from Pair-wise Comparisons. To be presented at NIPS, 2012 in Lake Tahoe. Full version available at Arxiv.

• A. Agarwal, S. Negahban, and M.J. Wainwright. Stochastic optimization and sparse statistical recovery: An optimal algorithm for high dimensions. To be presented at NIPS, 2012 in Lake Tahoe. Full version available at Arxiv.

• S. Negahban and D. Shah. Learning Sparse Boolean Polynomials. Presented at Allerton, 2012.

• S. Negahban, B.I.P. Rubinstein, and J. Gemmell. Scaling Multiple-Source Entity Resolution using Statistically Efficient Transfer Learning. To be presented at CIKM, 2012.

• A. Agarwal, S. Negahban, and M. J. Wainwright. (2012) Noisy matrix decomposition via convex relaxation: Optimal rates in high dimensions. Annals of Statistics, 40(2): 1171--1197, April, 2012.

  Full version available at Arxiv.

• A. Agarwal, S. Negahban, and M. J. Wainwright. Fast global convergence of gradient methods for high-dimensional statistical recovery. ARXIV To appear in Annals of Statistics. Presented in part at NIPS Conference, December 2010.

• S. Negahban and M. J. Wainwright, Restricted strong convexity and weighted matrix completion: Optimal bounds with noise. [PDF]. Journal of Machine Learning Research, 13: 1665-1697, May 2012.

  Preliminary Versions

  ARXIV preprint, September 2010.
  

  Presented in part at NIPS Workshops, Vancouver, Canada, December 2010.

• S. Negahban and M. J. Wainwright. Estimation of (near) low-rank matrices with noise and high-dimensional scaling. Appeared in part at ICML 2010, Haifa, Israel. [PDF]

  Journal Version:

  S. Negahban and M. J. Wainwright (2011). Estimation of (near) low-rank matrices with noise and high-dimensional scaling. Annals of Statistics, Vol 39, Number 2, pp. 1069--1097. [PDF] and Supplementary material.

  Preliminary Versions:

  To appear in Annals of Statistics.
  

  Preliminary Verison: Arxiv Paper
  

• Joint support recovery under high-dimensional scaling: Benefits and perils of $\ell_{1,\infty}$-regularization. S. Negahban and M. J. Wainwright. Advances in Neural Information Processing Systems, December 2008. Vancouver, Canada. [PDF]

  Journal version:

  S. Negahban and M. J. Wainwright (2011), Simultaneous support recovery in high dimensions: Benefits and perils of block $\ell_1/\ell_\infty$-regularization. IEEE Transactions on Information Theory, 57(6):3841--3863, June 2011.

  Preliminary version:

  S. Negahban and M. J. Wainwright. Simultaneous support recovery in high dimensions: Benefits and perils of block $\ell_1/\ell_\infty$-regularization. UC Berkeley Technical Report 774, May 2009. [Full PDF]
  

Reading Group

We have a reading group that meets once a week and discusses various papers involving statistics and computation. The reading group has a webpage that can be found at Reading Group.

Teaching

In the Fall of 2008 I was a Graduate Student Instructor for CS 281A, Berkeley's course on statistical learning theory and graphical models. In the Fall of 2010, I was a Graduate Student Instructor for EECS20n, Structure and Interpretation of Signals and Systems.

Courses

Measure Theory and Functional Analysis, Information Theory and Statistics, Probability Theory, Advanced Statistical Learning Theory, Statistical Learning Theory, Topology and Measure Theory, Statistical Signal Processing, Large Scale Convex Optimization, Convex Optimization, Advanced Topics in Information Theory, Mathematical Statistics