Learning-with-Label-Noise

A curated list of resources for Learning with Noisy Labels

Papers

• 2008-NIPS - Whose vote should count more: Optimal integration of labels from labelers of unknown expertise. [Paper][Code]

• 2009-ICML - Supervised learning from multiple experts: whom to trust when everyone lies a bit. [Paper]

• 2012-ICML - Learning to Label Aerial Images from Noisy Data. [Paper]

• 2013-NIPS - Learning with Multiple Labels. [Paper]

• 2013-NIPS - Learning with Noisy Labels. [Paper][Code]

• 2014-ML - Learning from multiple annotators with varying expertise. [Paper]

• 2014 - A Comprehensive Introduction to Label Noise. [Paper]

• 2014-Survey - Classification in the Presence of Label Noise: a Survey. [Paper]

• 2014 - Learning from Noisy Labels with Deep Neural Networks. [Paper]

• 2015-ICLR_W - Training Convolutional Networks with Noisy Labels. [Paper][Code]

• 2015-CVPR - Learning from Massive Noisy Labeled Data for Image Classification. [Paper][Code]

• 2015-CVPR - Visual recognition by learning from web data: A weakly supervised domain generalization approach. [Paper][Code]

• 2015-CVPR - Training Deep Neural Networks on Noisy Labels with Bootstrapping. [Paper][Loss-Code-Unofficial-1][Loss-Code-Unofficial-2][Code-Keras]

• 2015-ICCV - Webly supervised learning of convolutional networks. [Paper][Project Pagee]

• 2015-TPAMI - Classification with noisy labels by importance reweighting. [Paper][Code]

• 2015-NIPS - Learning with Symmetric Label Noise: The Importance of Being Unhinged. [Paper][Loss-Code-Unofficial]

• 2015 - Making Risk Minimization Tolerant to Label Noise. [Paper]

• 2015 - Learning Discriminative Reconstructions for Unsupervised Outlier Removal. [Paper][Code]

• 2015-TNLS - Rboost: label noise-robust boosting algorithm based on a nonconvex loss function and the numerically stable base learners. [Paper]

• 2016-AAAI - Robust semi-supervised learning through label aggregation. [Paper]

• 2016-ICLR - Auxiliary Image Regularization for Deep CNNs with Noisy Labels. [Paper][Code]

• 2016-CVPR - Seeing through the Human Reporting Bias: Visual Classifiers from Noisy Human-Centric Labels. [Paper][Code]

• 2016-ICML - Loss factorization, weakly supervised learning and label noise robustness. [Paper]

• 2016-RL - On the convergence of a family of robust losses for stochastic gradient descent. [Paper]

• 2016-NC - Noise detection in the Meta-Learning Level. [Paper]

• 2016-ECCV - The Unreasonable Effectiveness of Noisy Data for Fine-Grained Recognition. [Paper][Project Page]

• 2016-ICASSP - Training deep neural-networks based on unreliable labels. [Paper][Poster][Code-Unofficial]

• 2016-ICDM - Learning deep networks from noisy labels with dropout regularization. [Paper][Code]

• 2016-KBS - A robust multi-class AdaBoost algorithm for mislabeled noisy data. [Paper]

• 2017-AAAI - Robust Loss Functions under Label Noise for Deep Neural Networks. [Paper]

• 2017-ICLR - Training deep neural-networks using a noise adaptation layer. [Paper][Code]

• 2017-ICLR - Who Said What: Modeling Individual Labelers Improves Classification. [Paper]

• 2017-CVPR - Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach. [Paper] [Code]

• 2017-CVPR - Learning From Noisy Large-Scale Datasets With Minimal Supervision. [Paper]

• 2017-CVPR - Lean crowdsourcing: Combining humans and machines in an online system. [Paper][Code]

• 2017-CVPR - Attend in groups: a weakly-supervised deep learning framework for learning from web data. [Paper][Code]

• 2017-ICML - Robust Probabilistic Modeling with Bayesian Data Reweighting. [Paper][Code]

• 2017-ICCV - Learning From Noisy Labels With Distillation. [Paper][Code]

• 2017-NIPS - Toward Robustness against Label Noise in Training Deep Discriminative Neural Networks. [Paper]

• 2017-NIPS - Active bias: Training more accurate neural networks by emphasizing high variance samples. [Paper][Code]

• 2017-NIPS - Decoupling" when to update" from" how to update". [Paper][Code]

• 2017-IEEE-TIFS - A Light CNN for Deep Face Representation with Noisy Labels. [Paper][Code-Pytorch][Code-Keras][Code-Tensorflow]

• 2017-TNLS - Improving Crowdsourced Label Quality Using Noise Correction. [Paper]

• 2017-ML - Learning to Learn from Weak Supervision by Full Supervision. [Paper][Code]

• 2017-ML - Avoiding your teacher's mistakes: Training neural networks with controlled weak supervision. [Paper]

• 2017 - Deep Learning is Robust to Massive Label Noise. [Paper]

• 2017 - Fidelity-weighted learning. [Paper]

• 2017 - Self-Error-Correcting Convolutional Neural Network for Learning with Noisy Labels. [Paper]

• 2017 - Learning with confident examples: Rank pruning for robust classification with noisy labels. [Paper]

• 2017 - Regularizing neural networks by penalizing confident output distributions. [Paper]

• 2017 - Learning with Auxiliary Less-Noisy Labels. [Paper]

• 2018-AAAI - Deep learning from crowds. [Paper]

• 2018-ICLR - mixup: Beyond Empirical Risk Minimization. [Paper] [Code]

• 2018-ICLR - Learning From Noisy Singly-labeled Data. [Paper] [Code]

• 2018-ICLR - Dimensionality Driven Learning for Noisy Labels. [Paper] [Code]

• 2018-ICLR_W - How Do Neural Networks Overcome Label Noise?. [Paper]

• 2018-CVPR - CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise. [Paper] [Code]

• 2018-CVPR - Joint Optimization Framework for Learning with Noisy Labels. [Paper] [Code][Code-Unofficial-Pytorch]

• 2018-CVPR - Iterative Learning with Open-set Noisy Labels. [Paper] [Code]

• 2018-ICML - MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels. [Paper] [Code]

• 2018-ICML - Learning to Reweight Examples for Robust Deep Learning. [Paper] [Code] [Code-Unofficial-PyTorch]

• 2018-ICML - Dimensionality-Driven Learning with Noisy Labels. [Paper] [Code]

• 2018-ECCV - CurriculumNet: Weakly Supervised Learning from Large-Scale Web Images. [Paper] [Code]

• 2018-ECCV - Deep Bilevel Learning. [Paper] [Code]

• 2018-ECCV - Learning with Biased Complementary Labels. [Paper]

• 2018-ISBI - Training a neural network based on unreliable human annotation of medical images. [Paper]

• 2018-WACV - Iterative Cross Learning on Noisy Labels. [Paper]

• 2018-WACV - A semi-supervised two-stage approach to learning from noisy labels. [Paper]

• 2018-NIPS - Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels. [Paper] [Code]

• 2018-NIPS - Masking: A New Perspective of Noisy Supervision. [Paper] [Code]

• 2018-NIPS - Using Trusted Data to Train Deep Networks on Labels Corrupted by Severe Noise. [Paper] [Code]

• 2018-NIPS - Robustness of conditional GANs to noisy labels. [Paper] [Code]

• 2018-NIPS - Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels. [Paper][Loss-Code-Unofficial]

• 2018-TIP - Deep learning from noisy image labels with quality embedding. [Paper]

• 2018-TNLS - Progressive Stochastic Learning for Noisy Labels. [Paper]

• 2018 - Multiclass Learning with Partially Corrupted Labels. [Paper]

• 2018 - Improving Multi-Person Pose Estimation using Label Correction. [Paper]

• 2018 - Robust Determinantal Generative Classifier for Noisy Labels and Adversarial Attacks. [Paper]

• 2019-AAAI - Safeguarded Dynamic Label Regression for Generalized Noisy Supervision. [Paper] [Code][Slides][Poster]

• 2019-AAAI - Safeguarded Dynamic Label Regression for Noisy Supervision. [Paper] [Code]

• 2019-ICLR_W - SOSELETO: A Unified Approach to Transfer Learning and Training with Noisy Labels.[Paper][Code]

• 2019-CVPR - Learning to Learn from Noisy Labeled Data. [Paper] [Code]

• 2019-CVPR - Learning a Deep ConvNet for Multi-label Classification with Partial Labels. [Paper]

• 2019-CVPR - Label-Noise Robust Generative Adversarial Networks. [Paper] [Code]

• 2019-CVPR - Learning From Noisy Labels By Regularized Estimation Of Annotator Confusion. [Paper][Code]

• 2019-CVPR - Probabilistic End-to-end Noise Correction for Learning with Noisy Labels. [Paper][Code]

• 2019-CVPR - Graph Convolutional Label Noise Cleaner: Train a Plug-and-play Action Classifier for Anomaly Detection. [Paper][Code]

• 2019-CVPR - Improving Semantic Segmentation via Video Propagation and Label Relaxation. [Paper][Code]

• 2019-CVPR - Devil is in the Edges: Learning Semantic Boundaries from Noisy Annotations. [Paper] [Code][Project-page]

• 2019-CVPR - Noise-Tolerant Paradigm for Training Face Recognition CNNs. [Paper] [Code]

• 2019-CVPR - A Nonlinear, Noise-aware, Quasi-clustering Approach to Learning Deep CNNs from Noisy Labels. [Paper]

• 2019-IJCAI - Learning Sound Events from Webly Labeled Data. [Paper] [Code]

• 2019-ICML - Unsupervised Label Noise Modeling and Loss Correction. [Paper] [Code]

• 2019-ICML - Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels. [Paper] [Code]

• 2019-ICML - How does Disagreement Help Generalization against Label Corruption?. [Paper] [Code]

• 2019-ICML - Using Pre-Training Can Improve Model Robustness and Uncertainty [Paper] [Code]

• 2019-ICML - On Symmetric Losses for Learning from Corrupted Labels [Paper] [Poster] [Slides] [Code]

• 2019-ICML - Combating Label Noise in Deep Learning Using Abstention [Paper] [Code]

• 2019-ICASSP - Learning Sound Event Classifiers from Web Audio with Noisy Labels. [Paper] [Code]

• 2019-TGRS - Hyperspectral Image Classification in the Presence of Noisy Labels. [Paper] [Code]

• 2019-ICCV - NLNL: Negative Learning for Noisy Labels. [Paper][Code]

• 2019-ICCV - Symmetric Cross Entropy for Robust Learning With Noisy Labels. [Paper][Code]

• 2019-ICCV - Co-Mining: Deep Face Recognition With Noisy Labels.[Paper]

• 2019-ICCV - O2U-Net: A Simple Noisy Label Detection Approach for Deep Neural Networks.[Paper]

• 2019-ICCV - Deep Self-Learning From Noisy Labels. [Paper]

• 2019-ICCV_W - Photometric Transformer Networks and Label Adjustment for Breast Density Prediction. [Paper]

• 2019-NIPS - Meta-Weight-Net: Learning an Explicit Mapping For Sample Weighting.[Paper][Code]

• 2019-TPAMI - Learning from Large-scale Noisy Web Data with Ubiquitous Reweighting for Image Classification. [Paper]

• 2019-ISBI - Robust Learning at Noisy Labeled Medical Images: Applied to Skin Lesion Classification. [Paper]

• 2019-NIPS - Are Anchor Points Really Indispensable in Label-Noise Learning?. [Paper][Code]

• 2019-NIPS - Noise-tolerant fair classification. [Paper][Code]

• 2019-NIPS - Correlated Uncertainty for Learning Dense Correspondences from Noisy Labels. [Paper]

• 2019-NIPS - Combinatorial Inference against Label Noise. [Paper][Code]

• 2019-NIPS - L_DMI: A Novel Information-theoretic Loss Function for Training Deep Nets Robust to Label Noise. [Paper][Code]

• 2019 - ChoiceNet: Robust Learning by Revealing Output Correlations. [Paper]

• 2019 - Robust Learning Under Label Noise With Iterative Noise-Filtering. [Paper]

• 2019 - Synthetic vs Real: Deep Learning on Controlled Noise. [Paper]

• 2019 - IMAE for Noise-Robust Learning: Mean Absolute Error Does Not Treat Examples Equally and Gradient Magnitude's Variance Matters. [Paper][Project page]

• 2019 - Confident Learning: Estimating Uncertainty in Dataset Labels. [Paper] [Code]

• 2019 - Derivative Manipulation for General Example Weighting. [Paper] [Code]

• 2019 - Towards Robust Learning with Different Label Noise Distributions. [Paper]

• 2020-AAAI - Reinforcement Learning with Perturbed Rewards. [Paper] [Code]

• 2020-AAAI - Less Is Better: Unweighted Data Subsampling via Influence Function. [Paper] [Code]

• 2020-AAAI - Label Error Correction and Generation Through Label Relationships. [Paper]

• 2020-AAAI - Self-Paced Robust Learning for Leveraging Clean Labels in Noisy Data. [Paper]

• 2020-AAAI - Coupled-view Deep Classifier Learning from Multiple Noisy Annotators. [Paper]

• 2020-AAAI - Partial Multi-label Learning with Noisy Label Identification. [Paper]

• 2020-WACV - A Novel Self-Supervised Re-labeling Approach for Training with Noisy Labels. [Paper]

• 2020-WACV - Disentangling Human Dynamics for Pedestrian Locomotion Forecasting with Noisy Supervision. [Paper]

• 2020-WACV - Learning from Noisy Labels via Discrepant Collaborative Training. [Paper]

• 2020-ICLR - SELF: Learning to Filter Noisy Labels with Self-Ensembling. [Paper]

• 2020-ICLR - DivideMix: Learning with Noisy Labels as Semi-supervised Learning. [Paper]

• 2020-ICLR - Can gradient clipping mitigate label noise?. [Paper]

• 2020-ICLR - Curriculum Loss: Robust Learning and Generalization against Label Corruption. [Paper]

• 2020-ICLR - Simple and Effective Regularization Methods for Training on Noisily Labeled Data with Generalization Guarantee. [Paper]

• 2020-CVPR - Combating noisy labels by agreement: A joint training method with co-regularization. [Paper]

• 2020-CVPR - Distilling Effective Supervision From Severe Label Noise. [Paper][Code]

• 2020-CVPR - Learning From Noisy Anchors for One-Stage Object Detection. [Paper]

• 2020-CVPR - Self-Training With Noisy Student Improves ImageNet Classification. [Paper]

• 2020-CVPR - Noise Robust Generative Adversarial Networks. [Paper]

• 2020-CVPR - Noise-Aware Fully Webly Supervised Object Detection. [Paper][Code]

• 2020-CVPR - Global-Local GCN: Large-Scale Label Noise Cleansing for Face Recognition. [Paper]

• 2020-ICML - Learning with Bounded Instance-and Label-dependent Label Noise. [Paper]

• 2020-ICML - Label-Noise Robust Domain Adaptation.

• 2020-ICML - LTF: A Label Transformation Framework for Correcting Label Shift.

• 2020-ICML - Does label smoothing mitigate label noise?. [Paper]

• 2020-ICML - Error-Bounded Correction of Noisy Labels.

• 2020-ICML - Deep k-NN for Noisy Labels. [Paper]

• 2020-ICML - Searching to Exploit Memorization Effect in Learning from Noisy Labels. [Paper]

• 2020-ICML - Beyond Synthetic Noise: Deep Learning on Controlled Noisy Labels.

• 2020-ICML - Peer Loss Functions: Learning from Noisy Labels without Knowing Noise Rates. [Paper]

• 2020-ICML - Improving Generalization by Controlling Label-Noise Information in Neural Network Weights. [Paper][Code]

• 2020-ICML - Learning Binary Neurons with Noisy Supervision.

• 2020-ICML - S2GA: Robust Deep Learning with Noisy Labels without Early Stopping.

• 2020-ICML - Normalized Loss Functions for Deep Learning with Noisy Labels.

• 2020 - Multi-Class Classification from Noisy-Similarity-Labeled Data. [Paper]

• 2020 - NoiseRank: Unsupervised Label Noise Reduction with Dependence Models. [Paper]

• 2020 - Learning Adaptive Loss for Robust Learning with Noisy Labels. [Paper]

• 2020 - Identifying Mislabeled Data using the Area Under the Margin Ranking. [Paper]

Survey

• 2014-TNLS - Classification in the Presence of Label Noise: a Survey. [Paper]

• 2019 - Image Classification with Deep Learning in the Presence of Noisy Labels: A Survey. [Paper]

• 2020 - Deep learning with noisy labels: exploring techniques and remedies in medical image analysis. [Paper]

Github

• Search 'Noisy Label' Results
• Noisy Labels with Jupyter Notebook
• Noisy Label Neural Network1-Tensorflow
• Noisy Label Neural Network2-Chainer
• Multi-tasking Learning With Unreliable Labels
• Keras-noisy-lables-finetune
• Light CNN for Deep Face Recognition, in Tensorflow
• Rankpruning
• Cleanlab: machine learning python package for learning with noisy labels and finding label errors in datasets
• Deep Learning with Label Noise

Others

• Deep Learning Package-Chainer Tutorial
• Paper-Semi-Supervised Learning Literature Survey
• Cross Validated-Classification with Noisy Labels
• A little talk on label noise

Acknowledgements

Some of the above contents are borrowed from Noisy-Labels-Problem-Collection