• Home
  • Search
  • A Parametrical Model for Instance-Dependent Label Noise.
  • Cite Icon11
  • https://doi.org/10.1109/tpami.2023.3301876Copy DOI Icon

A Parametrical Model for Instance-Dependent Label Noise.

  • Abstract
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

In label-noise learning, estimating the transition matrix is a hot topic as the matrix plays an important role in building statistically consistent classifiers. Traditionally, the transition from clean labels to noisy labels (i.e., clean-label transition matrix (CLTM)) has been widely exploited on class-dependent label-noise (wherein all samples in a clean class share the same label transition matrix) to learn a clean-label classifier by employing the noisy data. However, the CLTM cannot handle the more common instance-dependent label-noise well (wherein the clean-to-noisy label transition matrix needs to be estimated at the instance level by considering the input quality) since the instance-dependent CLTM estimation requires to collect a set of clean labels from the noisy data distribution, which is difficult to achieve because the clean labels have uncertainty. Motivated by the fact that classifiers mostly output Bayes optimal labels for prediction, in this paper, we study to directly model the transition from Bayes optimal labels to noisy labels (i.e., Bayes-Label Transition Matrix (BLTM)) and learn a classifier to predict Bayes optimal labels. Note that given only noisy data, it is ill-posed to estimate either the CLTM or the BLTM. But favorably, Bayes optimal labels have no uncertainty compared with the clean labels, i.e., the class posteriors of Bayes optimal labels are one-hot vectors while those of clean labels are not. This enables two advantages to estimate the BLTM, i.e., (a) a set of examples with theoretically guaranteed Bayes optimal labels can be collected out of noisy data; (b) the feasible solution space is much smaller. By exploiting the advantages, this work proposes a parametrical model for estimating the instance-dependent label-noise transition matrix by employing a deep neural network, leading to better generalization and superior classification performance. From the theoretical perspective, we prove that by leveraging the instance-dependent Bayes-Label Transition Matrix, the classifier learned on the noisy data distribution would converge to the Bayes optimal classifier defined on the clean data distribution with an optimal parametric convergence rate for the empirical risk minimization.

Similar Papers
  • Research Article
  • Citations2

Solution Space Analysis and Feasible Genetic Algorithm for Assembly Job-Shop Scheduling Problems

  • Aug 01, 2009
  • Materials Science Forum
  • G.K Zhao +3
  • Research Article
  • Citations29

Optimizer convergence and local minima errors and their clinical importance

  • Aug 12, 2003
  • Physics in Medicine & Biology
  • Robert Jeraj +2
  • Research Article
  • Citations10

Label-Noise Robust Deep Generative Model for Semi-Supervised Learning

  • Jun 06, 2022
  • Technometrics
  • Heegeon Yoon +1
  • Conference Article

On-line scheduling of cluster tools based on hybrid quantum evolutionary algorithm

  • Aug 01, 2015
  • Linying Li +1
  • Conference Article
  • Citations3

On-line scheduling of robotic cells with post-processing residency constraints

  • Nov 10, 2003
  • Hyun Joong Yoon +1
  • Research Article
  • Citations68

Noise detection in the meta-learning level

  • May 08, 2015
  • Neurocomputing
  • Luís P.F Garcia +2
  • Supplementary Content

Nonparametric Perspective of Deep Learning

  • Dec 10, 2020
  • Figshare
  • Tianyang Hu
  • Research Article
  • Citations5

An optimization algorithm inspired by musical composition in constrained optimization problems

  • Aug 29, 2013
  • Revista de Matemática: Teoría y Aplicaciones
  • Roman Anselmo Mora-Gutiérrez +5
  • Research Article

Synergistic design method for MRE isolator combining extended-variable tri-phase topology and multi-objective parameter optimization

  • Oct 01, 2025
  • Smart Materials and Structures
  • Mi Zhu +5
  • Research Article

Classification and Feasibility Assessment of Infinitely Many Iso-Impulse Three-Dimensional Trajectories

  • Sep 26, 2025
  • The Journal of the Astronautical Sciences
  • Keziban Şaloğlu +1
  • Research Article
  • Citations43

Role of negative emissions technologies (NETs) and innovative technologies in transition of Japan’s energy systems toward net-zero CO2 emissions

  • Jan 30, 2021
  • Sustainability Science
  • Etsushi Kato +1
  • Research Article
  • Citations29

A multi-objective cooperation search algorithm for cascade reservoirs operation optimization considering power generation and ecological flows

  • Nov 23, 2023
  • Applied Soft Computing
  • Zhong-Kai Feng +4
  • Conference Article
  • Citations28

Variable-Topology Shape Optimization With a Control on Perimeter

  • Sep 11, 1994
  • Robert B Haber +2
  • Research Article

Exploring quantum annealing for coarse-grained protein folding.

  • Apr 08, 2026
  • Scientific reports
  • Timon Scheiber +2
  • Research Article
  • Citations1

Agent Negotiation in Water Policy Planning

  • Jan 01, 2011
  • International Journal of Digital Library Systems
  • Menq-Wen Lin +2
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.