Adaptive online prediction by following the perturbed leader

MaRDI QIDQ3093271zbMATH OpenFDO

Authors Marcus Hutter, Jan Poland

Publication date 12 October 2011

Full work available at URL http://www.jmlr.org/papers/v6/hutter05a.html

zbMATH Keywords

adaptive learning rateprediction with expert adviceadaptive adversaryonline sequential predictiongeneral weightsfollow the perturbed leaderexpected and high probability boundsgeneral alphabet and losshierarchy of experts

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05)

Recommendations

• Algorithmic Learning Theory
• Online learning in case of unbounded losses using follow the perturbed leader algorithm
• Adaptive and self-confident on-line learning algorithms
• The Follow Perturbed Leader Algorithm Protected from Unbounded One-Step Losses
• Follow the leader if you can, hedge if you must

Cited in

(14)

• Combining initial segments of lists
• A philosophical treatise of universal induction
• The Follow Perturbed Leader Algorithm Protected from Unbounded One-Step Losses
• Adaptive and self-confident on-line learning algorithms
• Follow the leader if you can, hedge if you must
• Following the Perturbed Leader to Gamble at Multi-armed Bandits
• Learning in auctions: regret is hard, envy is easy
• Algorithmic Learning Theory
• Stochastic Algorithms: Foundations and Applications
• Nonstochastic bandits: Countable decision set, unbounded costs and reactive environments
• The weak aggregating algorithm and weak mixability
• Online multiple kernel classification
• Online learning in case of unbounded losses using follow the perturbed leader algorithm
• Consistency of discrete Bayesian learning

This page was built for publication: Adaptive online prediction by following the perturbed leader

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3093271)