Pith. sign in

REVIEW 5 cited by

A Survey of Label-noise Representation Learning: Past, Present and Future

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2011.04406 v2 pith:CZK6TGR4 submitted 2020-11-09 cs.LG

classification cs.LG
keywords lnrllearningdirectionslabelsdeepmethodsmodelsnoisy
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Classical machine learning implicitly assumes that labels of the training data are sampled from a clean distribution, which can be too restrictive for real-world scenarios. However, statistical-learning-based methods may not train deep learning models robustly with these noisy labels. Therefore, it is urgent to design Label-Noise Representation Learning (LNRL) methods for robustly training deep models with noisy labels. To fully understand LNRL, we conduct a survey study. We first clarify a formal definition for LNRL from the perspective of machine learning. Then, via the lens of learning theory and empirical study, we figure out why noisy labels affect deep models' performance. Based on the theoretical guidance, we categorize different LNRL methods into three directions. Under this unified taxonomy, we provide a thorough discussion of the pros and cons of different categories. More importantly, we summarize the essential components of robust LNRL, which can spark new directions. Lastly, we propose possible research directions within LNRL, such as new datasets, instance-dependent LNRL, and adversarial LNRL. We also envision potential directions beyond LNRL, such as learning with feature-noise, preference-noise, domain-noise, similarity-noise, graph-noise and demonstration-noise.

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. FedGSCA: Medical Federated Learning with Global Sample Selector and Client Adaptive Adjuster under Label Noise

    cs.LG 2025-07 conditional novelty 6.0 of 10

    FedGSCA aggregates client-level GMM noise selectors and uses adaptive pseudo-labels with a credal-set robust loss to improve federated medical image classification under label noise.

  2. GFLC: Graph-based Fairness-aware Label Correction for Fair Classification

    cs.LG 2025-06 conditional novelty 6.0 of 10

    GFLC is a new label-correction method that uses confidence scores, graph curvature, and demographic parity to improve both accuracy and fairness under group-dependent label noise.

  3. Robust Losses from Univariate Base Functions for Noisy-Label Learning

    cs.LG 2026-07 conditional novelty 5.0 of 10

    Robust multiclass losses can be generated from univariate base functions by target-separation or pairwise binary-reduction mappings with derivative-based sufficient conditions.

  4. $\epsilon$-Softmax: Approximating One-Hot Vectors for Mitigating Label Noise

    cs.LG 2025-08 reject novelty 5.0 of 10

    ε-softmax, which adds a constant to the largest softmax probability and renormalizes, is claimed to make any loss noise-tolerant, but the required δ-condition is false for cross-entropy, so the theoretical promise is not met.

  5. Learning Causality for Modern Machine Learning

    cs.LG 2025-06 conditional novelty 2.0 of 10

    A thesis compiling six papers that use causal invariance to improve graph neural networks' out-of-distribution generalization, interpretability, and robustness.

Pith tools