Quantifying accuracy of learning via sample width

Martin Anthony, Joel Ratsaby

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

Abstract

In a recent paper, the authors introduced the notion of sample width for binary classifiers defined on the set of real numbers. It was shown that the performance of such classifiers could be quantified in terms of this sample width. This paper considers how to adapt the idea of sample width so that it can be applied in cases where the classifiers are defined on some finite metric space. We discuss how to employ a greedy set-covering heuristic to bound generalization error. Then, by relating the learning problem to one involving certain graph-theoretic parameters, we obtain generalization error bounds that depend on the sample width and on measures of 'density' of the underlying metric space.

Original languageEnglish
Title of host publicationProceedings of the 2013 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2013 - 2013 IEEE Symposium Series on Computational Intelligence, SSCI 2013
Pages84-90
Number of pages7
DOIs
StatePublished - 2013
Event2013 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2013 - 2013 IEEE Symposium Series on Computational Intelligence, SSCI 2013 - Singapore, Singapore
Duration: 16 Apr 201319 Apr 2013

Publication series

NameProceedings of the 2013 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2013 - 2013 IEEE Symposium Series on Computational Intelligence, SSCI 2013

Conference

Conference2013 IEEE Symposium on Foundations of Computational Intelligence, FOCI 2013 - 2013 IEEE Symposium Series on Computational Intelligence, SSCI 2013
Country/TerritorySingapore
CitySingapore
Period16/04/1319/04/13

Fingerprint

Dive into the research topics of 'Quantifying accuracy of learning via sample width'. Together they form a unique fingerprint.

Cite this