Learning with noisy labels
139 papers with code • 20 benchmarks • 16 datasets
Learning with noisy labels means When we say "noisy labels," we mean that an adversary has intentionally messed up the labels, which would have come from a "clean" distribution otherwise. This setting can also be used to cast learning from only positive and unlabeled data.
Benchmarks
These leaderboards are used to track progress in Learning with noisy labels
| Trend | Dataset | Best Model | Paper | Code | Compare |
|---|---|---|---|---|---|
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
|||||
|
Libraries
Use these libraries to find Learning with noisy labels models and implementationsDatasets
CIFAR-10
CIFAR-100
Food-101
VoxCeleb1
Clothing1M
WebVision
CIFAR-10N
Chaoyang
ANIMAL
Most implemented papers
Sharpness-Aware Minimization for Efficiently Improving Generalization
google-research/sam
•
•
ICLR 2021
In today's heavily overparameterized models, the value of the training loss provides few guarantees on model generalization ability.
Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels
bhanML/Co-teaching
•
•
NeurIPS 2018
Deep learning with noisy labels is practically challenging, as the capacity of deep models is so high that they can totally memorize these noisy labels sooner or later during training.
Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels
AlanChou/Truncated-Loss
•
•
NeurIPS 2018
Here, we present a theoretically grounded set of noise-robust loss functions that can be seen as a generalization of MAE and CCE.
Symmetric Cross Entropy for Robust Learning with Noisy Labels
YisenWang/symmetric_cross_entropy_for_noisy_labels
•
•
ICCV 2019
In this paper, we show that DNN learning with Cross Entropy (CE) exhibits overfitting to noisy labels on some classes ("easy" classes), but more surprisingly, it also suffers from significant under learning on some other classes ("hard" classes).
Confident Learning: Estimating Uncertainty in Dataset Labels
Confident learning (CL) is an alternative approach which focuses instead on label quality by characterizing and identifying label errors in datasets, based on the principles of pruning noisy data, counting with probabilistic thresholds to estimate noise, and ranking examples to train with confidence.
Normalized Loss Functions for Deep Learning with Noisy Labels
HanxunH/Active-Passive-Losses
•
•
ICML 2020
However, in practice, simply being robust is not sufficient for a loss function to train accurate DNNs.
Open-set Label Noise Can Improve Robustness Against Inherent Label Noise
hongxin001/ODNL
•
•
NeurIPS 2021
Learning with noisy labels is a practically challenging problem in weakly supervised learning.
Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotations
ucsc-real/cifar-10-100n
•
•
ICLR 2022
These observations require us to rethink the treatment of noisy labels, and we hope the availability of these two datasets would facilitate the development and evaluation of future learning with noisy label solutions.
Protoformer: Embedding Prototypes for Transformers
ashfarhangi/Protoformer
•
•
PAKDD 2022: Advances in Knowledge Discovery and Data Mining 2022
This paper proposes Protoformer, a novel self-learning framework for Transformers that can leverage problematic samples for text classification.
How does Disagreement Help Generalization against Label Corruption?
Learning with noisy labels is one of the hottest problems in weakly-supervised learning.
