ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Noise tolerance under risk minimization

Manwani, Naresh and Sastry, PS (2013) Noise tolerance under risk minimization. In: IEEE Transactions on Cybernetics, 43 (3). pp. 1146-1151.

[img] PDF
IEEE_Tran_Cyb_43-3_1146_2013.pdf - Published Version
Restricted to Registered users only

Download (261kB) | Request a copy
Official URL: http://dx.doi.org/10.1109/TSMCB.2012.2223460

Abstract

In this paper, we explore noise-tolerant learning of classifiers. We formulate the problem as follows. We assume that there is an unobservable training set that is noise free. The actual training set given to the learning algorithm is obtained from this ideal data set by corrupting the class label of each example. The probability that the class label of an example is corrupted is a function of the feature vector of the example. This would account for most kinds of noisy data one encounters in practice. We say that a learning method is noise tolerant if the classifiers learnt with noise-free data and with noisy data, both have the same classification accuracy on the noise-free data. In this paper, we analyze the noise-tolerance properties of risk minimization (under different loss functions). We show that risk minimization under 0-1 loss function has impressive noise-tolerance properties and that under squared error loss is tolerant only to uniform noise; risk minimization under other loss functions is not noise tolerant. We conclude this paper with some discussion on the implications of these theoretical results.

Item Type: Journal Article
Additional Information: Copyright of this article belongs to IEEE-Inst Electrical Electronics Engineers Inc.
Keywords: Label Noise; Loss Functions; Noise Tolerance; Risk Minimization
Department/Centre: Division of Electrical Sciences > Electrical Engineering
Depositing User: Francis Jayakanth
Date Deposited: 26 Jun 2013 12:29
Last Modified: 26 Jun 2013 12:29
URI: http://eprints.iisc.ac.in/id/eprint/46746

Actions (login required)

View Item View Item