Skip to content
conftrace_
2008 JMLR JMLR 2008

Nearly Uniform Validation Improves Compression-Based Error Bounds

Abstract

This paper develops bounds on out-of-sample error rates for support vector machines (SVMs). The bounds are based on the numbers of support vectors in the SVMs rather than on VC dimension. The bounds developed here improve on support vector counting bounds derived using Littlestone and Warmuth's compression-based bounding technique. [abs] [ pdf ][ bib ] © JMLR 2008. (edit, beta)

🐣 Hot Topic Early Bird - generalization bound

Authors

Eric Bax

Topics

Machine Learning > Core Methods > Classification Machine Learning > Optimization & Theory > Learning Theory

Keywords

vc dimension support vector machine generalization bound compression-based bound
Download PDF

More from JMLR 2008

All JMLR 2008 papers →
A New Algorithm for Estimating the Effective Dimension-Reduction Subspace
Value Function Based Reinforcement Learning in Changing Markovian Environments
Regularization on Graphs with Function-adapted Diffusion Processes
Learning from Multiple Sources
Exponentiated Gradient Algorithms for Conditional Random Fields and Max-Margin Markov Networks
Classification with a Reject Option using a Hinge Loss