Before joining IBM in Aug 2007 I spent 5 very enjoyable years at Harvard University as a PhD student advised by Leslie Valiant and as a postdoc.
My research interests are primarily in Machine Learning Theory and Computational Complexity. I'm particularly interested in questions in the overlap of these areas. I also work on understanding of natural learning systems: learning by the brain and evolution as learning. This work is based on the models pioneered by Leslie Valiant (brain, evolvability).
- Dec 2015: I am organizing a workshop on adaptive data analysis at NIPS 2015 (with Moritz Hardt, Aaron Roth and Adam Smith).
- July 2015: instructor at Berkeley summer course in mining and modeling of neuroscience data (with Christos Papadimitriou).
- March-May 2015: visiting scientist at Simons Institute for the Theory of Computing, UC Berkeley. Information Theory.
- March-May 2014: visiting scientist at Simons Institute for the Theory of Computing, UC Berkeley. Evolutionary Biology and the Theory of Computing.
- Sep-Oct 2013: visiting researcher at LIAFA, Universite Paris Diderot.
Recent/upcoming conference program committees:
- COLT 2016: Co-chair (with Sasha Rakhlin).
- AJCAI-ML 2015, ALT 2015, COLT 2015, ITCS 2015 , NIPS 2014, ALT 2014, COLT 2014 (publication chair), ALT 2013, COLT 2013.
Some recent talks:
- Preserving statistical validity in adaptive data analysis. @STOC 2015: slides.
- Approximate resilience, monotonicity, and the complexity of agnostic learning. @SODA 2015: slides.
- Sample complexity bounds on differentially private learning via communication complexity. @COLT 2014 and ITA 2015: slides.
- Using data privacy for better adaptive predictions. Foundations of Learning Theory workshop @COLT 2014 : slides.
- On the power and the limits of evolvability. Simons Institute workshop on Computational Theories of Evolution: slides.
- Optimal bounds on approximation of submodular and XOS functions by juntas. Simons Institute workshop on Real Analysis at @FOCS 2013 : slides.
- The reusable holdout: Preserving validity in adaptive data analysis.
With Cynthia Dwork, Moritz Hardt, Toniann Pitassi, Omer Reingold and Aaron Roth. Science, 2015.
Based on STOC and NIPS papers below. See also my post on this work at IBM Research blog.
Supplemental Material with some corrections can be found here.
- Generalization in Adaptive Data Analysis and Holdout Reuse.
With Cynthia Dwork, Moritz Hardt, Toniann Pitassi, Omer Reingold and Aaron Roth. NIPS, 2015 (to appear) .
- Preserving Statistical Validity in Adaptive Data Analysis.
With Cynthia Dwork, Moritz Hardt, Toniann Pitassi, Omer Reingold and Aaron Roth. STOC 2015 . Invited to Comm. of the ACM, Invited to SICOMP special issue on STOC.
- Subsampled Power Iteration: a Unified Algorithm for Block Models and Planted CSP's .
With Will Perkins and Santosh Vempala. NIPS 2015 (to appear) .
- Tight Bounds on Low-degree Spectral Concentration of Submodular and XOS Functions.
With Jan Vondrak. FOCS 2015 (to appear) .
- On the Complexity of Random Satisfiability Problems with Planted Solutions .
With Will Perkins and Santosh Vempala. STOC 2015 .
- Nearly Tight Bounds on ℓ1 Approximation of Self-Bounding Functions.
With Pravesh Kothari and Jan Vondrak. Preprint. Apr, 2014 .
- Agnostic Learning of Disjunctions on Symmetric Distributions.
With Pravesh Kothari. JMLR 2015 (to appear) .
- Approximate resilience, monotonicity, and the complexity of agnostic learning .
With Dana Dachman-Soled, Li-Yang Tan, Andrew Wan and Karl Wimmer. SODA, 2015 .
- The Statistical Query Complexity of Learning Sparse Halfspaces.
Open Problem at COLT 2014.
- Learning Coverage Functions and Private Release of Marginals.
With Pravesh Kothari. COLT 2014.
- Sample Complexity Bounds on Differentially Private Learning via Communication Complexity .
With David Xiao. COLT 2014, SICOMP 2015 (to appear).
- Statistical Active Learning Algorithms for Noise Tolerance and Differential Privacy.
With Nina Balcan. NIPS 2013 . Algorithmica. Special Issue on New Theoretical Challenges in Machine Learning (by invitation)
- Optimal Bounds on Approximation of Submodular and XOS Functions by Juntas.
With Jan Vondrak. FOCS 2013 . SICOMP 2015 (to appear), Special Issue on FOCS (by invitation), SIGecom Exchanges 12.2 (invited letter)
- Learning using Local Membership Queries.
With Pranjal Awasthi and Varun Kanade. COLT 2013, Best Student (Co-authored) Paper Award .
- Representation, Approximation and Learning of Submodular Functions Using Low-rank Decision Trees.
With Pravesh Kothari and Jan Vondrak. COLT 2013 .
- Provenance-based Dictionary Refinement in Information Extraction.
With Sudeepa Roy, Laura Chiticariu, Frederick Reiss and Huaiyu Zhu. SIGMOD 2013 .
- Statistical Algorithms and a Lower Bound for Detecting Planted Cliques.
With Elena Grigorescu, Lev Reyzin, Santosh Vempala and Ying Xiao. STOC 2013 . Full version updated in 2015.
- Computational Bounds on Statistical Query Learning.
With Varun Kanade. COLT 2012.
- Learning DNF Expressions from Fourier Spectrum.
- Nearly Optimal Solutions for the Chow Parameters Problem and Low-weight Approximation of Halfspaces.
With Anindya De, Ilias Diakonikolas and Rocco Servedio. STOC 2012; JACM 2014 .
- Distribution-Independent Evolvability of Linear Threshold Functions.
- Lower Bounds and Hardness Amplification for Learning Shallow Monotone Formulas.
With Homin Lee and Rocco Servedio. COLT 2011.
- Distribution-Specific Agnostic Boosting.
ITCS (formerly ICS) 2010.
- Agnostic Learning of Monomials by Halfspaces is Hard.
With Venkatesan Guruswami, Prasad Raghavendra and Yi Wu. FOCS 2009; SICOMP 2012 .
- A Complete Characterization of Statistical Query Learning with Applications to Evolvability.
FOCS 2009; JCSS 2012 (by invitation).
- Sorting and Selection with Imprecise Comparisons.
With Miklos Ajtai, Avinatan Hassidim and Jelani Nelson. ICALP A, 2009; TALG 2015
- Robustness of Evolvability.
- Experience-Induced Neural Circuits That Achieve High Capacity..
With Leslie Valiant. Neural Computation 21:10, 2009.
- On The Power of Membership Queries in Agnostic Learning.
COLT 2008; JMLR 10, 2009
- The Learning Power of Evolution.
With Leslie Valiant. Open Problems/New Directions at COLT 2008.
- Evolvability from Learning Algorithms.
- Separating Models of Learning with Faulty Teachers.
With Shrenik Shah. ALT 2007; TCS 410(19), 2009 (by invitation)
- New Results for Learning Noisy Parities and Halfspaces.
With Parikshit Gopalan, Subhash Khot, and Ashok Ponnuswami. FOCS 2006; SICOMP 39(2), 2009 (by invitation)
- Optimal Hardness Results for Maximizing Agreements with Monomials.
CCC 2006; SICOMP 39(2), 2009 (by invitation)
- Hardness of Approximate Two-level Logic Minimization and PAC Learning with Membership Queries.
STOC 2006; JCSS 75(1), 2009 (by invitation)
- Attribute Efficient and Non-adaptive Learning of Parities and DNF Expressions.
COLT 2005, Best Student Paper Award; JMLR 8, 2007 (by invitation)
- The Complexity of Properly Learning Simple Concept Classes.
With Misha Alekhnovich, Mark Braverman, Adam Klivans, and Toni Pitassi.
FOCS 2004; JCSS 74(1), 2008 (by invitation)
Ph.D. thesis. Efficiency and Computational Limitations of Learning Algorithms. Harvard University. January 2007
- Hardness of Proper Learning.
The Encyclopedia of Algorithms. Springer-Verlag, 2015
- Statistical Query Learning.
The Encyclopedia of Algorithms. Springer-Verlag, 2015
A bit more about me
Some of my favorite hobbies are ballroom dancing, mountain biking, photography, hiking, skiing, drinking tea and visiting Paris.