TY - GEN
T1 - Maximum margin multiclass nearest neighbors
AU - Kontorovich, Aryeh
AU - Weiss, Roi
N1 - Publisher Copyright:
Copyright © (2014) by the International Machine Learning Society (IMLS) All rights reserved.
PY - 2014
Y1 - 2014
N2 - 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.
AB - 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.
UR - http://www.scopus.com/inward/record.url?scp=84919949537&partnerID=8YFLogxK
M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???
AN - SCOPUS:84919949537
T3 - 31st International Conference on Machine Learning, ICML 2014
SP - 2501
EP - 2511
BT - 31st International Conference on Machine Learning, ICML 2014
T2 - 31st International Conference on Machine Learning, ICML 2014
Y2 - 21 June 2014 through 26 June 2014
ER -