Research Areas
Machine learning; sequential prediction; optimization; statistical learning theory; applied probability
Current Projects
Developing efficient and optimal algorithms for online learning with limited feedback. Providing minimax guarantees for sequential optimization. Studying the interplay of statistical accuracy and optimization complexity. Developing methods for efficient regularization through random perturbation.
Representative Publications
(with J. Abernethy)
"Beating the Adaptive Bandit with High Probability." Information Theory and Applications Workshop, forthcoming 2009.
(with J. Abernethy and E. Hazan)
"Competing in the Dark: An Efficient Algorithm for Bandit Linear Optimization." Conference on Learning Theory, 2008.
(with Jacob Abernethy, Peter Bartlett, and Ambuj Tewari)
"Optimal Strategies and Minimax Lower Bounds for Online Convex Games." Conference on Learning Theory, 2008.
(with Jacob Abernethy and Peter Bartlett)
"Multitask Learning with Expert Advice." Conference on Learning Theory, 2007.
(with Andrea Caponnetto)
"Stability Properties of Empirical Risk Minimization over Donsker Classes." Journal of Machine Learning Research Vol. 7, 2565-2583, (December 2006).
(with Dmitry Panchenko and Sayan Mukherjee)
"Risk Bounds for Mixture Density Estimation." ESAIM Probability and Statistics Vol. 9, 220-229, (June 2005).
Education
PhD, MIT 2006; BA, Cornell University, 2000
Academic Positions Held
Wharton: 2009-present. Previous Appointments: University of California, Berkeley
