List your company
ResearchFormerly WizWand

Image Classification on CIFAR10 (test) (Test Accuracy)

99Test AccuracyPanda et al.
79.281684.400889.5294.6392Jun 18, 2020Mar 11, 2021Dec 2, 2021Aug 26, 2022May 19, 2023Feb 9, 2024Nov 2, 2024
Updated 8mo ago

Evaluation results

MethodLinks
Panda et al.Model=beitv2, Public Data=ImageNet, epsilon=12023.0699
Mehta et al. [50]Model=ViT-G/16, Public Data=JFT, epsilon=82023.0698.84
Mehta et al. [50]Model=ViT-G/16, Public Data=JFT, epsilon=42023.0698.83
Mehta et al. [50]Model=ViT-G/16, Public Data=JFT, epsilon=12023.0698.8
Mehta et al. [50]Model=ViT-G/16, Public Data=JFT, epsilon=22023.0698.8
ProxylessNASParam (M)=5.7, Search Cost (GPU days)=4, Search Method=gradient2020.0697.92
BYOLEvaluation protocol=Fine-tuned, Backbone=ResNet-50, Pre-trained=ImageNet2022.0197.8
SimCLREvaluation protocol=Fine-tuned, Backbone=ResNet-50, Pre-trained=ImageNet2022.0197.7
ReLICv2Evaluation protocol=Fine-tuned, Backbone=ResNet-50, Pre-trained=ImageNet2022.0197.7
PDARTSParam (M)=3.4, Latency (ms)=40.9, Search Cost (GPU days)=0.3, Search Method=gradient, Search space=DARTS search space2020.0697.5
Supervised-INEvaluation protocol=Fine-tuned, Backbone=ResNet-50, Pre-trained=ImageNet2022.0197.5
MIMOConv staticN=1, # inputs/pass=12023.1297.49
AmoebaNet-BParam (M)=2.8, Search Cost (GPU days)=3150, Search Method=evolution2020.0697.45
NASNet-AParam (M)=3.3, Search Cost (GPU days)=1800, Search Method=RL2020.0697.35
WideIsoNet-28-10# inputs/pass=12023.1297.31
MIMOConv dynamic (N=1-4)# inputs/pass=12023.1297.13
ENASParam (M)=4.6, Search Cost (GPU days)=0.5, Search Method=RL2020.0697.11
GDASParam (M)=3.4, Latency (ms)=30.6, Search Cost (GPU days)=4, Search Method=gradient2020.0697.07
MIMOConv staticN=2, # inputs/pass=22023.1296.93
NAONetParam (M)=10.6, Search Cost (GPU days)=200, Search Method=NAO2020.0696.82
WideResNet-28-10# inputs/pass=12023.1296.82
PNASParam (M)=3.2, Search Cost (GPU days)=225, Search Method=SMBO2020.0696.59
DenseNet-BCParam (M)=25.6, Search Method=manual2020.0696.52
ResNeXt-29, 16x64dParam (M)=68.1, Search Method=manual2020.0696.42
MIMOConv dynamic (N=1-4)# inputs/pass=22023.1296.41
Wide ResNetParam (M)=36.5, Search Method=manual2020.0696.2
Random InitEvaluation protocol=Fine-tuned, Backbone=ResNet-50, Pre-trained=None2022.0195.9
MIMOConv staticN=4, # inputs/pass=42023.1295.58
MIMOConv dynamic (N=1-4)# inputs/pass=42023.1295.43
Mehta et al. [51]Model=ViT-B/16, Public Data=ImageNet, epsilon=82023.0695.2
Mehta et al. [51]Model=ViT-B/16, Public Data=ImageNet, epsilon=12023.0695.1
Mehta et al. [51]Model=ViT-B/16, Public Data=ImageNet, epsilon=22023.0695.1
Mehta et al. [51]Model=ViT-B/16, Public Data=ImageNet, epsilon=42023.0695.1
Π-net-ResNetBackbone=ResNet34, #p x 10^6=132021.0494.9
ResNet34Backbone=ResNet34, #p x 10^6=21.32021.0494.8
PDCBackbone=ResNet34, #p x 10^6=10.52021.0494.8
SENetBackbone=ResNet18, #p x 10^6=11.52021.0494.6
PDCBackbone=ResNet18, #p x 10^6=82021.0494.6
ResNet18Backbone=ResNet18, #p x 10^6=11.22021.0494.5
Π-net-ResNetBackbone=ResNet18, #p x 10^6=62021.0494.5
PDC-compBackbone=ResNet18, #p x 10^6=4.32021.0494.5
OrMo-DABackbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0794.5
OrMoBackbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0794.32
OrMoBackbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0794.26
shifted momentumBackbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0794.2
OrMo-DABackbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0794.03
shifted momentumBackbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0794.02
naive ASGDmBackbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0793.74
SMEGA*Backbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0793.72
NNCLREvaluation protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2022.0193.7
Supervised-INEvaluation protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2022.0193.6
SMEGA*Backbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0793.36
naive ASGDmBackbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0793.1
ReLICv2Evaluation protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2022.0192.8
OTTERBackbone=ViT-B/16, Setting=Zero-shot2024.0491.7
ASGDBackbone=ResNet18, Number of workers=8, Worker setting=heterogeneous2024.0791.52
ASGDBackbone=ResNet18, Number of workers=8, Worker setting=homogeneous2024.0791.45
BYOLEvaluation protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2022.0191.3
Prior MatchingBackbone=ViT-B/16, Setting=Zero-shot2024.0491.3
Implicit ETFNetwork=VGG13, Epochs=2002024.1190.98
SimCLREvaluation protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2022.0190.6
ResNet18-EntropyNumber of Samples=100002022.1090.4
FR-ResNet18-EntropyNumber of Samples=100002022.1090.36
StandardNetwork=VGG13, Epochs=2002024.1190.34
ResNet18-core-setNumber of Samples=100002022.1089.86
ResNet18-EntropyNumber of Samples=90002022.1089.83
LLALNumber of Samples=100002022.1089.78
FR-ResNet18-EntropyNumber of Samples=90002022.1089.59
LLALNumber of Samples=90002022.1089.18
FR-ResNet18-EntropyNumber of Samples=80002022.1088.91
ResNet18-core-setNumber of Samples=90002022.1088.67
LLALNumber of Samples=80002022.1088.55
ResNet18-EntropyNumber of Samples=80002022.1088.33
Zero-shotBackbone=ViT-B/16, Setting=Zero-shot2024.0488.3
Implicit ETFNetwork=VGG13, Epochs=502024.1188.3
FR-ResNet18-EntropyNumber of Samples=70002022.1087.77
ResNet18-core-setNumber of Samples=80002022.1087.56
LLALNumber of Samples=70002022.1087.4
ResNet18-EntropyNumber of Samples=70002022.1087.05
StandardNetwork=VGG13, Epochs=502024.1186.7
FR-ResNet18-EntropyNumber of Samples=60002022.1086.33
ResNet18-core-setNumber of Samples=70002022.1086.24
LLALNumber of Samples=60002022.1085.95
DP-RandPModel=WRN-16-4, Public Data=Synthetic, epsilon=82023.0685.26
ResNet18-EntropyNumber of Samples=60002022.1085.13
Implicit ETFNetwork=ResNet18, Epochs=2002024.1184.78
Fixed ETFNetwork=ResNet18, Epochs=2002024.1184.53
LLALNumber of Samples=50002022.1084.14
FR-ResNet18-EntropyNumber of Samples=50002022.1084.1
ResNet18-core-setNumber of Samples=60002022.1084.04
DP-RandPModel=WRN-16-4, Public Data=Synthetic, epsilon=62023.0684.01
StandardNetwork=ResNet18, Epochs=2002024.1183.97
DP-RandPModel=WRN-16-4, Public Data=Synthetic, epsilon=42023.0681.88
FR-ResNet18-EntropyNumber of Samples=40002022.1081.86
Implicit ETFNetwork=ResNet18, Epochs=502024.1181.76
ResNet18-EntropyNumber of Samples=50002022.1081.27
LLALNumber of Samples=40002022.1080.83
Fixed ETFNetwork=ResNet18, Epochs=502024.1180.63
StandardNetwork=ResNet18, Epochs=502024.1180.47
ResNet18-core-setNumber of Samples=50002022.1080.04