References & Citations
Computer Science > Machine Learning
Title: Learning from Binary Labels with Instance-Dependent Corruption
(Submitted on 3 May 2016 (v1), last revised 4 May 2016 (this version, v2))
Abstract: Suppose we have a sample of instances paired with binary labels corrupted by arbitrary instance- and label-dependent noise. With sufficiently many such samples, can we optimally classify and rank instances with respect to the noise-free distribution? We provide a theoretical analysis of this question, with three main contributions. First, we prove that for instance-dependent noise, any algorithm that is consistent for classification on the noisy distribution is also consistent on the clean distribution. Second, we prove that for a broad class of instance- and label-dependent noise, a similar consistency result holds for the area under the ROC curve. Third, for the latter noise model, when the noise-free class-probability function belongs to the generalised linear model family, we show that the Isotron can efficiently and provably learn from the corrupted sample.
Submission history
From: Aditya Menon [view email][v1] Tue, 3 May 2016 04:47:02 GMT (364kb,D)
[v2] Wed, 4 May 2016 04:59:21 GMT (361kb,D)
Link back to: arXiv, form interface, contact.