Machine Learning

, Volume 69, Issue 2, pp 115–142

A primal-dual perspective of online learning algorithms

Authors

    • School of Computer Science & EngineeringThe Hebrew University
  • Yoram Singer
    • School of Computer Science & EngineeringThe Hebrew University
    • Google Inc.
Article

DOI: 10.1007/s10994-007-5014-x

Cite this article as:
Shalev-Shwartz, S. & Singer, Y. Mach Learn (2007) 69: 115. doi:10.1007/s10994-007-5014-x

Abstract

We describe a novel framework for the design and analysis of online learning algorithms based on the notion of duality in constrained optimization. We cast a sub-family of universal online bounds as an optimization problem. Using the weak duality theorem we reduce the process of online learning to the task of incrementally increasing the dual objective function. The amount by which the dual increases serves as a new and natural notion of progress for analyzing online learning algorithms. We are thus able to tie the primal objective value and the number of prediction mistakes using the increase in the dual.

Keywords

Online learningMistake boundsDualityRegret bounds
Download to read the full article text

Copyright information

© Springer Science+Business Media, LLC 2007