论文信息 - Localization and Adaptation in Online Learning

Localization and Adaptation in Online Learning

We introduce a formalism of localization for online learning problems, which, similarly to statistical learning theory, can be used to obtain fast rates. In particular, we introduce local sequential Rademacher complexities and other local measures. Based on the idea of relaxations for deriving algorithms, we provide a template method that takes advantage of localization. Furthermore, we build a general adaptive method that can take advantage of the suboptimality of the observed sequence. We illustrate the utility of the introduced concepts on several problems. Among them is a novel upper bound on regret in terms of classical Rademacher complexity when the data are i.i.d.

[1] Ambuj Tewari,et al. Online Learning: Random Averages, Combinatorial Parameters, and Learnability , 2010, NIPS.

[2] Elad Hazan,et al. Extracting certainty from uncertainty: regret bounded by variation in costs , 2008, Machine Learning.

[3] Peter L. Bartlett,et al. A Stochastic View of Optimal Regret through Minimax Duality , 2009, COLT.

[4] V. Koltchinskii,et al. Empirical margin distributions and bounding the generalization error of combined classifiers , 2002, math/0405343.

[5] Wouter M. Koolen,et al. Adaptive Hedge , 2011, NIPS.

[6] Ohad Shamir,et al. Relax and Randomize : From Value to Algorithms , 2012, NIPS.

[7] Peter L. Bartlett,et al. Adaptive Online Gradient Descent , 2007, NIPS.

[8] Yoav Freund,et al. A Parameter-free Hedging Algorithm , 2009, NIPS.

[9] Ambuj Tewari,et al. Online Learning: Stochastic, Constrained, and Smoothed Adversaries , 2011, NIPS.

[10] Gábor Lugosi,et al. Prediction, learning, and games , 2006 .

[11] P. Bartlett,et al. Local Rademacher complexities , 2005, math/0508275.

[12] Elad Hazan,et al. Logarithmic regret algorithms for online convex optimization , 2006, Machine Learning.