2011/10/28 by Tim van Erven, Peter Grünwald, Wouter M. Koolen +1 · 1 citation
Mathematics · #stat.ML
published as Advances in Neural Information Processing Systems 24, pages 1656-1664, December 2011 · This is the full version of the paper with the same name that will appear in Advances in Neural Information Processing Systems 24 (NIPS 2011), 2012. The two papers are identical, except that this version contains an extra section of Additional Material
arxiv created 2011/10/28 · arxiv updated 2015/03/04
Most methods for decision-theoretic online learning are based on the Hedge algorithm, which takes a parameter called the learning rate. In most previous analyses the learning rate was carefully tuned to obtain optimal worst-case performance, leading to suboptimal performance on easy instances, for example when there exists an action that is significantly better than all others. We propose a new way of setting the learning rate, which adapts to the difficulty of the learning problem: in the worst case our procedure still guarantees optimal performance, but on easy instances it achieves much smaller regret. In particular, our adaptive method achieves constant regret in a probabilistic setting, when there exists an action that on average obtains strictly smaller loss than all other actions. We also provide a simulation study comparing our approach to existing methods.