Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Imprecise Label Learning: A Unified Framework for Learning with Various Imprecise Label Configurations

Hao Chen, Ankit Shah, Jindong Wang, Ran Tao, Yidong Wang, Xing Xie, Masashi Sugiyama, Rita Singh, Bhiksha Raj

TL;DR

Imprecise label learning (ILL) is introduced, a framework for the unification of learning with various imprecise label configurations, marking the first unified framework with robust and effective performance across various challenging settings.

Abstract

Learning with reduced labeling standards, such as noisy label, partial label, and multiple label candidates, which we generically refer to as \textit{imprecise} labels, is a commonplace challenge in machine learning tasks. Previous methods tend to propose specific designs for every emerging imprecise label configuration, which is usually unsustainable when multiple configurations of imprecision coexist. In this paper, we introduce imprecise label learning (ILL), a framework for the unification of learning with various imprecise label configurations. ILL leverages expectation-maximization (EM) for modeling the imprecise label information, treating the precise labels as latent variables.Instead of approximating the correct labels for training, it considers the entire distribution of all possible labeling entailed by the imprecise information. We demonstrate that ILL can seamlessly adapt to partial label learning, semi-supervised learning, noisy label learning, and, more importantly, a mixture of these settings. Notably, ILL surpasses the existing specified techniques for handling imprecise labels, marking the first unified framework with robust and effective performance across various challenging settings. We hope our work will inspire further research on this topic, unleashing the full potential of ILL in wider scenarios where precise labels are expensive and complicated to obtain.

Imprecise Label Learning: A Unified Framework for Learning with Various Imprecise Label Configurations

TL;DR

Imprecise label learning (ILL) is introduced, a framework for the unification of learning with various imprecise label configurations, marking the first unified framework with robust and effective performance across various challenging settings.

Abstract

Learning with reduced labeling standards, such as noisy label, partial label, and multiple label candidates, which we generically refer to as \textit{imprecise} labels, is a commonplace challenge in machine learning tasks. Previous methods tend to propose specific designs for every emerging imprecise label configuration, which is usually unsustainable when multiple configurations of imprecision coexist. In this paper, we introduce imprecise label learning (ILL), a framework for the unification of learning with various imprecise label configurations. ILL leverages expectation-maximization (EM) for modeling the imprecise label information, treating the precise labels as latent variables.Instead of approximating the correct labels for training, it considers the entire distribution of all possible labeling entailed by the imprecise information. We demonstrate that ILL can seamlessly adapt to partial label learning, semi-supervised learning, noisy label learning, and, more importantly, a mixture of these settings. Notably, ILL surpasses the existing specified techniques for handling imprecise labels, marking the first unified framework with robust and effective performance across various challenging settings. We hope our work will inspire further research on this topic, unleashing the full potential of ILL in wider scenarios where precise labels are expensive and complicated to obtain.
Paper Structure (35 sections, 22 equations, 2 figures, 21 tables)

This paper contains 35 sections, 22 equations, 2 figures, 21 tables.

Table of Contents

  1. Introduction
  2. Preliminary
  3. Imprecise Label Learning
  4. A Unified Framework for Learning with Imprecise Labels
  5. Instantiating the Unified EM Formulation
  6. Experiments
  7. Partial Label Learning
  8. Semi-Supervised Learning
  9. Noisy Label Learning
  10. Mixed Imprecise Label Learning
  11. Conclusion
  12. Notation
  13. Related Work
  14. Methods
  15. Derivation of Variational Lower Bound
  16. ...and 20 more sections

Figures (2)

  • Figure 1: Illustration of the full label and imprecise label configurations. We use an example dataset of 4 training instances and 3 classes. (a) Full label, the annotation is a single true label; (b) Partial label, the annotation is a label candidate set containing true label; (c) Semi-supervised, only part of the dataset is labeled, and the others are unlabeled; (d) Noisy label, the annotation is mislabeled.
  • Figure 2: Baseline model pipelines for various imprecise label configurations. (a) PiCO wang2022pico for partial label learning. (b) FixMatch sohn2020fixmatch for semi-supervised learning. (c) SOP sopliu22w for noisy label learning. (d) The proposed unified framework. It accommodates any imprecise label configurations and also mixed imprecise labels with an EM formulation.