Maximum margin multiclass nearest neighbors

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We develop a general framework for margin- based multicategory classification in metric spaces. The basic work-horse is a margin- regularized version of the nearest-neighbor classifier. We prove generalization bounds that match the state of the art in sample size n and significantly improve the dependence on the number of classes κ. Our point of departure is a nearly Bayes-optimal finite-sample risk bound independent of κ. Although κ-free, this bound is un- regularized and non-adaptive, which motivates our main result: Rademacher and scale-sensitive margin bounds with a logarithmic dependence on κ. As the best previous risk estimates in this setting were of order √κ, our bound is exponentially sharper. From the algorithmic standpoint, in doubling metric spaces our classifier may be trained on n examples in CJ(n2 log n) time and evaluated on new points in 0(log n) time.

Original languageAmerican English
Title of host publication31st International Conference on Machine Learning, ICML 2014
Pages2501-2511
Number of pages11
ISBN (Electronic)9781634393973
StatePublished - 1 Jan 2014
Event31st International Conference on Machine Learning, ICML 2014 - Beijing, China
Duration: 21 Jun 201426 Jun 2014

Publication series

Name31st International Conference on Machine Learning, ICML 2014
Volume3

Conference

Conference31st International Conference on Machine Learning, ICML 2014
Country/TerritoryChina
CityBeijing
Period21/06/1426/06/14

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Computer Networks and Communications
  • Software

Fingerprint

Dive into the research topics of 'Maximum margin multiclass nearest neighbors'. Together they form a unique fingerprint.

Cite this