PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Convex Repeated Games and Fenchel Duality
Shai Shalev-Shwartz and Yoram Singer
In: NIPS 2006, 5-10 Dec 2006, Vancouver, CA.

There is a more recent version of this eprint available. Click here to view it.

Abstract

We describe and analyze an algorithmic framework for playing convex repeated games. In each trial of the repeated game, the first player predicts a vector and then the second player responds with a loss function for the vector. We describe a generalization of Fenchel duality, which is used to derive an algorithmic framework for the first player and analyze the player's regret. We then use our algorithmic framework and its corresponding regret analysis for analyzing online learning and boosting algorithms.

EPrint Type:Conference or Workshop Item (Spotlight)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
Theory & Algorithms
ID Code:2523
Deposited By:Shai Shalev-Shwartz
Deposited On:22 November 2006

Available Versions of this Item