AI Paper Atlas

Research map / Benchmarks / CIFAR-100

CIFAR-100 leaderboard

1,173 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on CIFAR-100. Best accuracy: 99.98 (MCSF, CVPR 2026).

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-100 · accuracy

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

#accuracy ↑MethodPaperTopic
199.98MCSFMulti-Hierarchical Contrastive Spectral Fusion for Multi-View Clustering
CVPR 2026 · Table 1
Clustering
299.2TrainProVeTraining Data Provenance Verification: Did Your Model Use Synthetic Data from My Generative Model for Training?
CVPR 2025 · Table 2
Text-to-image diffusion
399.2Proposed label recovery (label smoothing)
ResNet50
Towards Eliminating Hard Label Constraints in Gradient Inversion Attacks
ICLR 2024 · Table: Experiments for label r
Recognition and long-tail
495.97GCFAggMVCGCFAgg: Global and Cross-View Feature Aggregation for Multi-View Clustering
CVPR 2023 · Table 3
Clustering
595.95ViT-Calibrator (+All) on Eva-L
Eva-L
ViT-Calibrator: Decision Stream Calibration for Vision Transformer
AAAI 2024 · Table 1
Attention and transformers
695.52CAPrompt
ImageNet21K ViT-B/16
CAPrompt: Cyclic Prompt Aggregation for Pre-Trained Model Based Class Incremental Learning
AAAI 2025 · Table 1
Domain and prompt-based
795.3pFedSeq
ViT-B/16 + LoRA
Look Back for More: Harnessing Historical Sequential Updates for Personalized Federated Adapter Tuning
AAAI 2025 · Table 1
Personalized and heterogeneous FL
894.86PRESTO
ResNet18 (reduced capacity)
Discrete Continuous Optimization Framework for Simultaneous Clustering and Training in Mixture Models
ICML 2023 · Table 2
Clustering
994.85HetDPT (15 layers)
DINO-V2-Giant
Rethinking Depth Pruning for Vision Transformers: A Heterogeneity-Aware Perspective
ICML 2026 · Table 6
Attention and transformers
1094.8RCNMCReliable Clustering Number Estimation for Contrastive Multi-View Clustering
CVPR 2026 · Table 2
Clustering
828 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-100 · error rate

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

#error rate ↓MethodPaperTopic
12.27SSWPL
WRN-28-2
Self-Supervised Wasserstein Pseudo-Labeling for Semi-Supervised Image Classification
CVPR 2021 · Table 1
Semi-supervised learning
23.54MC-CP
CNN
Robust Uncertainty Quantification Using Conformalised Monte Carlo Prediction
AAAI 2024 · Table 1
Uncertainty and Gaussian processes
34.192MFABA
ResNet-50
MFABA: A More Faithful and Accelerated Boundary-Based Attribution Method for Deep Neural Networks
AAAI 2024 · Attribution Error Rate Results
Prediction and decision losses
49.01IHDA+
PyramidNet
Post-training Iterative Hierarchical Data Augmentation for Deep Networks
NeurIPS 2020 · Table 2
Recognition and long-tail
510.7PyramidNet+ShakeDrop + AutoAugment
PyramidNet+ShakeDrop
AutoAugment: Learning Augmentation Strategies From Data
CVPR 2019 · Table 2
Neural architecture search
610.94PBA
PyramidNet+ShakeDrop
Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
ICML 2019 · Table 2
Visual and representation RL
711.2DADA
PyramidNet+ShakeDrop
DADA: Differentiable Automatic Data Augmentation
ECCV 2020 · Table 2
Neural architecture search
811.9Proposed
PyramidNet+ShakeDrop
Learning Augmentation Network via Influence Functions
CVPR 2020 · Table 2
Recognition and long-tail
912.24CutMix + SLA+SD
PyramidNet200
Self-supervised Label Augmentation via Input Transformations
ICML 2020 · Table 4
Self-supervised and contrastive
1013.5rSGD
PyramidNet+ShakeDrop
Entropic gradient descent algorithms and wide flat minima
ICLR 2021 · Table 1
Implicit bias and generalization
167 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-100 · top-1 error

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

#top-1 error ↓MethodPaperTopic
110.48MCMC-Aug
PyramidNet
A Continuous Mapping For Augmentation Design
NeurIPS 2021 · Table 3
Optimizers
213.05Saliency Grafting + ShakeDrop
PyramidNet-200
Saliency Grafting: Innocuous Attribution-Guided Mixup with Calibrated Label Mixing
AAAI 2022 · Table: PyramidNet-200 CIFAR-10
Recognition and long-tail
313.47PyramidNet-200 + CutMix + ShakeDrop + MoEx
PyramidNet-200
On Feature Normalization and Data Augmentation
CVPR 2021 · Table 2
Autoregressive and latent generation
413.81CutMix + ShakeDrop
PyramidNet-200
CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features
ICCV 2019 · Table 4
Weakly supervised detection and segmentation
514.17Pyramid-200 + StyleCutMix (auto-γ)
PyramidNet-200
StyleMix: Separating Content and Style for Enhanced Data Augmentation
CVPR 2021 · Table 3
Certified and empirical robustness
615.49PS-KD
PyramidNet
Self-Knowledge Distillation With Progressive Refinement of Targets
ICCV 2021 · Table 1
Knowledge distillation
715.54RM
DenseNet-161
RecursiveMix: Mixed Learning with History
NeurIPS 2022 · Table 1
CNN features
816.3OnlineAugment + AutoAugment
Wide-ResNet-28-10
OnlineAugment: Online Data Augmentation with Less Domain Knowledge
ECCV 2020 · Table 6
Domain adaptation and generalization
916.35PyramidNet bottleneck, α=200Deep Pyramidal Residual Networks
CVPR 2017 · Table: CIFAR results
CNN design
1016.528PS-ResNeXt-29 (16x64d)
ResNeXt-29 (16x64d)
PSConv: Squeezing Feature Pyramid into One Compact Poly-Scale Convolutional Layer
ECCV 2020 · Table 6
CNN design
33 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-100 · clean accuracy

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

#clean accuracy ↑MethodPaperTopic
181.27Co-training, radius 8/255
ViT-S
Revisiting adapters with adversarial training
ICLR 2023 · Table: Evaluating on other dat
Certified and empirical robustness
279.65SL
WRN-16-10
Understanding Square Loss in Training Overparametrized Neural Network Classifiers
NeurIPS 2022 · Table 1
Kernels and regression theory
377.2VIPAug
ResNet-18
Domain Generalization with Vital Phase Augmentation
AAAI 2024 · CIFAR-10 and CIFAR-100 compari
Domain adaptation and generalization
476.25RoLI - Full-FT
Swin Transformer
Initialization Matters for Adversarial Transfer Learning
CVPR 2024 · Table 2
Certified and empirical robustness
576.2PubDef
WideResNet-34-10
PubDef: Defending Against Transfer Attacks From Public Models
ICLR 2024 · Table, Clean and adversarial a
Black-box and transfer attacks
676Ours
ResNet18
Generalization Bound and New Algorithm for Clean-Label Backdoor Attack
ICML 2024 · Evaluations on more datasets
Backdoor and poisoning
774.81AVmixup
WRN-34-10
Adversarial Vertex Mixup: Toward Better Adversarially Robust Generalization
CVPR 2020 · Table: comparisons on other da
Certified and empirical robustness
873.16IGAM-MNIST
MNIST robust teacher
What It Thinks Is Important Is Important: Robustness Transfers Through Input Gradients
CVPR 2020 · Table 3
Certified and empirical robustness
972.2Noisy mixup
ResNet-18
Boundary thickness and robustness in learning models
NeurIPS 2020 · Table 1
Certified and empirical robustness
1071.72OCM256 WRN-28-4 (8-bit)
WRN-28-4
Improving Robustness Against Stealthy Weight Bit-Flip Attacks by Output Code Matching
CVPR 2022 · Table 2
Attacks on text and graphs
21 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-100 · ECE

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

19 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-100 · accuracy (val)

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

10 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-100 · FID

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

#FID ↓MethodPaperTopic
12.72CBDM
DDPM
Class-Balancing Diffusion Models
CVPR 2023 · Table 1
Fast diffusion sampling
22.86CORAL
DDPM
CORAL: Disentangling Latent Representations in Long-Tailed Diffusion
NeurIPS 2025 · Balanced dataset table
Recognition and long-tail
33.86OursDenoising Likelihood Score Matching for Conditional Score-based Data Generation
ICLR 2022 · Table: evaluation results on C
Fast diffusion sampling
44.88EGC
U-Net
EGC: Image Generation and Classification via a Diffusion Energy-Based Model
ICCV 2023 · Table 2
Fast diffusion sampling
55.77BigGAN + MC + DiffAug (Hinge)
BigGAN
MCGAN: Enhancing GAN Training with Regression-Based Generator Loss
AAAI 2025 · Table: Quantitative results of
GANs
67.15FQ-TAC-GAN
TAC-GAN
Feature Quantization Improves GAN Training
ICML 2020 · CIFAR-100 comparison table
Recognition and long-tail
77.22TAC-GAN
SN-GAN
Twin Auxilary Classifiers GAN
NeurIPS 2019 · Table 1
GANs
88.05BigGAN Hinge SAN*
BigGAN
SAN: Inducing Metrizability of GAN with Discriminative Normalized Linear Layer
ICLR 2024 · Table: FID and IS results with
GANs
98.12ADC-GANConditional GANs with Auxiliary Discriminative Classifier
ICML 2022 · Table: FID and Intra-FID and A
Recognition and long-tail
109.2SSAH
ResNet-20
Frequency-Driven Imperceptible Adversarial Attack on Semantic Similarity
CVPR 2022 · Table 1
Attacks on text and graphs
9 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-100 · AUROC

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

#AUROC ↑MethodPaperTopic
198.86CSL
ResNet18
Towards Memorization Estimation: Fast, Formal and Free
ICML 2025 · Table 3
Prediction and decision losses
298.1TrainProVeTraining Data Provenance Verification: Did Your Model Use Synthetic Data from My Generative Model for Training?
CVPR 2025 · Table 2
Text-to-image diffusion
398.1Ours
ViT-B_16
Fake It Until You Make It : Towards Accurate Near-Distribution Novelty Detection
ICLR 2023 · Table 1(a)
Open-set and few-shot detection
493.6UNODE
WideResNet-50-2 pretrained on ImageNet
Universal Novelty Detection Through Adaptive Contrastive Learning
CVPR 2024 · Table 1
OOD and anomaly detection
593.49Curv ZO NLL
ResNet18
Curvature Clues: Decoding Deep Learning Privacy with Input Loss Curvature
NeurIPS 2024 · Table 1
Certified and empirical robustness
693.14T.S. (w/ AS, multiple)
ViT-B/16
Great Models Think Alike: Improving Model Reliability via Inter-Model Latent Agreement
ICML 2023 · Table 1
Molecules and proteins
792.48R2D2-Net (SVGI)
AlexNet
Feature Preserving Shrinkage on Bayesian Neural Networks Via the R2D2 Prior
TPAMI 2025 · Table 3
Uncertainty and Gaussian processes
892TeCo
PreActResNet18
Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistency
CVPR 2023 · Table: evaluation results
Backdoor and poisoning
990.74SCALE
DenseNet-101
Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhancement
ICLR 2024 · Table: OOD detection results o
OOD and anomaly detection
1090ETP
LeNet5
Evidential Turing Processes
ICLR 2022 · Experiments table
Domain adaptation and generalization
8 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-100 · NLL

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

8 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-100 · top-5 accuracy

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

#top-5 accuracy ↑MethodPaperTopic
195.79MLP-Mixer (ReLU)
gMLP-S
Unraveling Attention via Convex Duality: Analysis and Interpretations of Vision Transformers
ICML 2022 · Table 1
Attention and transformers
295.41DiffuseMix
PreActResNet-18
DiffuseMix: Label-Preserving Data Augmentation with Diffusion Models
CVPR 2024 · Table 2
Fast diffusion sampling
394.28STATA
Spikingformer
Towards Efficient Spiking Transformer: a Token Sparsification Framework for Training and Inference Acceleration
ICML 2024 · Table 5
Spiking networks
494RCNet (RCAS, searched on CIFAR-100)Resource Constrained Neural Network Architecture Search: Will a Submodularity Assumption Help?
ICCV 2019 · Table 2
Neural architecture search
592.62Anchor loss
ResNet-110
Anchor Loss: Modulating Loss Scale Based on Prediction Difficulty
ICCV 2019 · Table 1
Recognition and long-tail
691.72E²-Train (SMD+SLU+PSG)
ResNet-110
E2-Train: Training State-of-the-art CNNs with Over 80% Energy Savings
NeurIPS 2019 · Table: ResNet-110 and MobileNe
CNN design
791.51Feedback NetFeedback Networks
CVPR 2017 · Endpoint performance compariso
RNNs and normalization
890.2EP, L=15, block size 2
ff-EBM
Towards training digitally-tied analog blocks via hybrid gradient computation
NeurIPS 2024 · Table
RNNs and normalization
989.6DPᵀ, α=0, β=1.0
VGG16
Two Tales of Single-Phase Contrastive Hebbian Learning
ICML 2024 · Table 3
RNNs and normalization
1088.36DP
VGG16
Dual Propagation: Accelerating Contrastive Hebbian Learning with Dyadic Neurons
ICML 2023 · CIFAR10, CIFAR100 and ImageNet
RNNs and normalization
4 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-100 · PGD-20 accuracy

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

#PGD-20 accuracy ↑MethodPaperTopic
148.05DCS + GSAT
WRN 34
Adversarial Robustness via Deformable Convolution with Stochasticity
ICML 2025 · Table 1
Certified and empirical robustness
247.28Feature Scatter + AVmixupAdversarial Vertex Mixup: Toward Better Adversarially Robust Generalization
CVPR 2020 · Table: Feature Scatter compari
Certified and empirical robustness
344.69TPAP
ResNet-18
Robust Overfitting Does Matter: Test-Time Adversarial Purification With FGSM
CVPR 2024 · Table 1
Certified and empirical robustness
444.33CTRW
WRN34
Adversarial Robustness through Random Weight Sampling
NeurIPS 2023 · Table 2
Certified and empirical robustness
544.07AT+S²O
PreAct ResNet18
Enhancing Adversarial Training With Second-Order Statistics of Weights
CVPR 2022 · Table: Adversarial training ac
Certified and empirical robustness
640.15SAFER
DeiT-Ti
SAFER: Sharpness Aware layer-selective Finetuning for Enhanced Robustness in vision transformers
ICCV 2025 · Table 1
Certified and empirical robustness
737.37RPF
ResNet-18
Adversarial Robustness via Random Projection Filters
CVPR 2023 · Table 1
Certified and empirical robustness
832.87IAD-I
WideResNet-34-10; TRADES teacher
Reliable Adversarial Distillation with Unreliable Teachers
ICLR 2022 · Table 2
Attacks on text and graphs
932.3DARWIN
WRN-34
Robust Distillation via Untargeted and Targeted Intermediate Adversarial Samples
CVPR 2024 · Table 1
Certified and empirical robustness
1030.6TRNASTRNAS: A Training-Free Robust Neural Architecture Search
ICCV 2025 · Table 3
Neural architecture search
4 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-100 · natural accuracy

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

#natural accuracy ↑MethodPaperTopic
178.8SAM (ρ=0.2)
PreActResNet-18
On the Duality Between Sharpness-Aware Minimization and Adversarial Training
ICML 2024 · Table: Natural and robust accu
Certified and empirical robustness
266.69DCS + GSAT
WRN 34
Adversarial Robustness via Deformable Convolution with Stochasticity
ICML 2025 · Table 1
Certified and empirical robustness
364.49AWD + ASAM, WRN32-10
WRN32-10
Improving Robustness with Adaptive Weight Decay
NeurIPS 2023 · Table 2
Certified and empirical robustness
462.97Generalist
ResNet-18
Generalist: Decoupling Natural and Robust Generalization
CVPR 2023 · Table 3
Certified and empirical robustness
562.56NADAR-B
WRN32-10
Neural Architecture Dilation for Adversarial Robustness
NeurIPS 2021 · Table: CIFAR-100 validation ac
Certified and empirical robustness
660.62CTRW
WRN34
Adversarial Robustness through Random Weight Sampling
NeurIPS 2023 · Table 2
Certified and empirical robustness
760.01UIAT (One-off)
ResNet-18
The Enemy of My Enemy Is My Friend: Exploring Inverse Adversaries for Improving Adversarial Training
CVPR 2023 · Table 1
Certified and empirical robustness
859.12DARWIN
WRN-34
Robust Distillation via Untargeted and Targeted Intermediate Adversarial Samples
CVPR 2024 · Table 1
Certified and empirical robustness
955.47AT WRN ARD → MNV2 (α=0.93)
AT WRN
Adversarially Robust Distillation
AAAI 2020 · Table 7
Certified and empirical robustness
1055.14WSCAT
WRN-28-10
Weakly Supervised Contrastive Adversarial Training for Learning Robust Features from Semi-supervised Data
CVPR 2025 · Table 3
Attacks on text and graphs
2 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

1 less reported leaderboard for CIFAR-100 (other metrics or splits) are in the CSV and the API.

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