AI Paper Atlas

Research map / Benchmarks / CIFAR-10 (4000 labels)

CIFAR-10 (4000 labels) leaderboard

52 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on CIFAR-10 (4000 labels). Best error rate: 3.38 (RegMixMatch, AAAI 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

CIFAR-10 (4000 labels) · error rate

41 papers, each with its best reported number for its own method; lower is better.

#error rate ↓MethodPaperTopic
13.38RegMixMatch
Wide ResNet-28-2
RegMixMatch: Optimizing Mixup Utilization in Semi-Supervised Learning
AAAI 2025 · Table 1
Semi-supervised learning
23.61FlatMatch
WRN-28-2
FlatMatch: Bridging Labeled Data and Unlabeled Data with Cross-Sharpness for Semi-Supervised Learning
NeurIPS 2023 · Table 1
Semi-supervised learning
33.8SemCo (μ=7)
WideResNet-28-2
All Labels Are Not Created Equal: Enhancing Semi-Supervision via Label Grouping and Co-Training
CVPR 2021 · Table 1
Semi-supervised learning
43.82GSF
WRN-28-2
Towards Discovering the Effectiveness of Moderately Confident Samples for Semi-Supervised Learning
CVPR 2022 · Table 1
Semi-supervised learning
53.84Multi-Head Co-TrainingSemi-supervised Learning with Multi-Head Co-Training
AAAI 2022 · Table 1
Semi-supervised learning
63.86STOCO
WRN-28-2
Stochastic Consensus: Enhancing Semi-Supervised Learning with Consistency of Stochastic Classifiers
ECCV 2022 · Table 2
Semi-supervised learning
73.88FixMatch + IAM-DInconsistency-Aware Minimization: Improving Generalization with Unlabeled Data
ICML 2026 · Table 3
Self-supervised and contrastive
83.92ConMatch-PConMatch: Semi-Supervised Learning with Confidence-Guided Consistency Regularization
ECCV 2022 · Table 1
Semi-supervised learning
94.1FreeMatch
WRN-28-2
FreeMatch: Self-adaptive Thresholding for Semi-supervised Learning
ICLR 2023 · Table 1
Semi-supervised learning
104.11NP-Match
WRN-28-2
NP-Match: When Neural Processes meet Semi-Supervised Learning
ICML 2022 · Table 1
Semi-supervised learning
31 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

CIFAR-10 (4000 labels) · accuracy

11 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
196.25FullMatch
FixMatch
Boosting Semi-Supervised Learning by Exploiting All Unlabeled Data
CVPR 2023 · Table 1
Semi-supervised learning
296.14FlexMatch (w/SAA)
FlexMatch, WRN-28-2
Enhancing Sample Utilization through Sample Adaptive Augmentation in Semi-Supervised Learning
ICCV 2023 · Table 1
Semi-supervised learning
395.94FixMatch (STS)
FixMatch
Spatial Ensemble: a Novel Model Smoothing Mechanism for Student-Teacher Framework
NeurIPS 2021 · Table 6
Knowledge distillation
495.78Ours (super-class discriminative learning)Improving Barely Supervised Learning by Discriminating Unlabeled Samples with Super-Class
NeurIPS 2022 · Table 3
Semi-supervised learning
595.57InstanT
WRN-28-2
InstanT: Semi-supervised Learning with Instance-dependent Thresholds
NeurIPS 2023 · Table 2
Semi-supervised learning
695.11FixMatch + LP-A3
FixMatch
Adversarial Auto-Augment with Label Preservation: A Representation Learning Principle Guided Approach
NeurIPS 2022 · Table 1
Recognition and long-tail
789.58UnMixMatchScaling Up Semi-supervised Learning with Unconstrained Unlabelled Data
AAAI 2024 · Table 1
Semi-supervised learning
888Our Model + Mean TeacherMitigating Embedding and Class Assignment Mismatch in Unsupervised Image Classification
ECCV 2020 · Table 4
Recognition and long-tail
981.79CNN with Semantic Loss
CNN
A Semantic Loss Function for Deep Learning with Symbolic Knowledge
ICML 2018 · CIFAR results table
Recognition and long-tail
1075.46GDEN-NLData Representation and Learning With Graph Diffusion-Embedding Networks
CVPR 2019 · Table 3
Graph self-supervised learning
1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.