Instance-Dependent Label-Noise Learning with Manifold-Regularized Transition Matrix Estimation

TL;DR

This paper introduces a manifold-regularized approach to estimate instance-dependent transition matrices in label-noise learning, leveraging geometric assumptions to improve stability and accuracy in noisy label scenarios.

Contribution

It proposes a novel manifold-based regularization method that effectively estimates instance-dependent transition matrices under label noise, addressing unidentifiability issues.

Findings

Outperforms state-of-the-art methods on synthetic datasets

Demonstrates robustness on real-world noisy datasets

Reduces estimation error without increasing approximation error

Abstract

In label-noise learning, estimating the transition matrix has attracted more and more attention as the matrix plays an important role in building statistically consistent classifiers. However, it is very challenging to estimate the transition matrix T(x), where x denotes the instance, because it is unidentifiable under the instance-dependent noise(IDN). To address this problem, we have noticed that, there are psychological and physiological evidences showing that we humans are more likely to annotate instances of similar appearances to the same classes, and thus poor-quality or ambiguous instances of similar appearances are easier to be mislabeled to the correlated or same noisy classes. Therefore, we propose assumption on the geometry of T(x) that "the closer two instances are, the more similar their corresponding transition matrices should be". More specifically, we formulate above assumption into the manifold embedding, to effectively reduce the degree of freedom of T(x) and make it stably estimable in practice. The proposed manifold-regularized technique works by directly reducing the estimation error without hurting the approximation error about the estimation problem of T(x). Experimental evaluations on four synthetic and two real-world datasets demonstrate that our method is superior to state-of-the-art approaches for label-noise learning under the challenging IDN.