List your company
ResearchFormerly WizWand

Image Classification on CIFAR-10 (test) (Error Rate)

1.99Error RateEnAET
0.68649.485718.28527.0843Mar 25, 2019Aug 8, 2019Dec 23, 2019May 8, 2020Sep 21, 2020Feb 5, 2021Jun 22, 2021
Updated 8mo ago

Evaluation results

MethodLinks
EnAETBackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.111.99
ProxylessNASBackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.112.08
PDARTSParam (M)=3.4, Architecture Optimization=gradient2021.062.5
P-DARTSSearch Time (GPU days)=0.32021.062.5
MdeNASParam (M)=3.6, Architecture Optimization=gradient2021.062.55
AABackbone=Wide-ResNet-28-102020.032.6
PBABackbone=Wide-ResNet-28-102020.032.6
OHL AABackbone=Wide-ResNet-28-102020.032.6
AutoAugmentBackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.112.6
PBABackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.112.6
NASNet-AParam (M)=3.3, Architecture Optimization=RL2021.062.65
NASNet-AParameters (M)=3.3, Search Time (GPU days)=18002021.062.65
SETNParam (M)=4.6, Architecture Optimization=gradient2021.062.69
Fast AABackbone=Wide-ResNet-28-102020.032.7
DADABackbone=Wide-ResNet-28-102020.032.7
Fast AABackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.112.7
CutMixBackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.112.88
ENASParam (M)=4.6, Architecture Optimization=RL2021.062.89
ENASParameters (M)=4.6, Search Time (GPU days)=0.52021.062.89
GDASParam (M)=3.4, Architecture Optimization=gradient2021.062.93
Fractional MP# Params=12.0M2019.063.47
PDO-eConvG (Equivariance Group)=p8, Depth=26, Number of Parameters=4.6M2020.073.5
EnAETBackbone=Wide ResNet-28-2, Evaluation Protocol=Fully Supervised2019.113.55
PDO-eConvG (Equivariance Group)=p8, Depth=44, Number of Parameters=2.62M2020.073.68
BaselineBackbone=Wide ResNet-28-2-Large, Supervision Type=Fully Supervised2019.113.9
Wide ResNetG (Equivariance Group)=Z2, Depth=26, Number of Parameters=36.5M2020.074
DenseNetk=12, # Params=7.0M2019.064.1
AutoAugmentBackbone=Wide ResNet-28-2, Evaluation Protocol=Fully Supervised2019.114.1
MixMatchBackbone=Wide ResNet-28-2, Evaluation Protocol=Fully Supervised2019.114.13
G-CNNG (Equivariance Group)=p4m, Depth=26, Number of Parameters=7.2M2020.074.17
Wide ResNet-40-4# Params=8.9M2019.064.53
FractalNet# Params=38.6M2019.064.6
ImageNet-32 Fine-tunedPre-training=ImageNet-32, Number of labels=50,0002019.064.61
ResNet-1001# Params=10.2M2019.064.62
SESEMI ASLArchitecture=ConvNet, # Params=3.1M2019.064.7
SESEMI SSLBackbone=ConvNet, Number of labels=50,0002019.064.7
SESEMI ASLArchitecture=WRN-28-2, # Params=1.5M2019.064.71
SESEMI SSLBackbone=WRN, Number of labels=50,0002019.064.71
ResNetG (Equivariance Group)=Z2, Depth=1001, Number of Parameters=10.3M2020.074.92
G-CNNG (Equivariance Group)=p4m, Depth=44, Number of Parameters=2.62M2020.074.94
BCN+WSModel=ResNet-18, Micro-batch=false2019.034.96
DNModel=ResNet-18, Micro-batch=false2019.035.02
BaselineBackbone=Wide ResNet-28-2, Evaluation Protocol=Fully Supervised2019.115.1
BNModel=ResNet-18, Micro-batch=false2019.035.2
EnAETNumber of labels=40002019.115.35
PDO-eConvG (Equivariance Group)=p6m, Depth=26, Number of Parameters=0.37M2020.075.38
BCN+WSModel=ResNet-18, Micro-batch=true2019.035.43
Pi Model SSLNumber of labels=50,0002019.065.56
TempEns SSLNumber of labels=50,0002019.065.6
SNModel=ResNet-18, Micro-batch=false2019.035.6
ResNetG (Equivariance Group)=Z2, Depth=44, Number of Parameters=2.64M2020.075.61
PDO-eConvG (Equivariance Group)=p6, Depth=26, Number of Parameters=0.36M2020.075.65
VAT SSLNumber of labels=50,0002019.065.81
SupervisedArchitecture=ConvNet, # Params=3.1M2019.065.82
SupervisedNumber of labels=50,0002019.065.82
FitResNetInitialization=LSUV, # Params=2.5M2019.065.84
Mean Teacher SSLNumber of labels=50,0002019.065.94
EnAETNumber of labels=20002019.116
EnAETNumber of labels=10002019.116.95
SE-RegNet-20(ConvLSTM)Layers=20, Regulator=ConvLSTM, Squeeze-and-Excitation=true2021.016.98
SE-RegNet-20(ConvGRU)Layers=20, Regulator=ConvGRU, Squeeze-and-Excitation=true2021.017.25
EnAETNumber of labels=5002019.117.27
RegNet-20(ConvLSTM)Layers=20, Regulator=ConvLSTM2021.017.28
RegNet-20(ConvGRU)Layers=20, Regulator=ConvGRU2021.017.42
DNModel=ResNet-18, Micro-batch=true2019.037.55
SE-RegNet-20(ConvRNN)Layers=20, Regulator=ConvRNN, Squeeze-and-Excitation=true2021.017.55
RegNet-20(ConvRNN)Layers=20, Regulator=ConvRNN2021.017.6
SNModel=ResNet-18, Micro-batch=true2019.037.62
Highway Network# Params=2.3M2019.067.72
SE-ResNet-20Layers=20, Squeeze-and-Excitation=true2021.018.02
ResNet-20Layers=202021.018.38
BNModel=ResNet-18, Micro-batch=true2019.038.45
HexaConvG (Equivariance Group)=p6m, Depth=26, Number of Parameters=0.34M2020.078.64
EnAETNumber of labels=1002019.119.35
HexaConvG (Equivariance Group)=p6, Depth=26, Number of Parameters=0.34M2020.079.98
ImageNet-32 Fine-tunedPre-training=ImageNet-32, Number of labels=4,0002019.0610.16
SESEMI SSLBackbone=WRN, Number of labels=4,0002019.0611.23
VAT SSLNumber of labels=1,0002019.0611.36
ResNetG (Equivariance Group)=Z2, Depth=26, Number of Parameters=0.37M2020.0711.5
SESEMI SSLBackbone=ConvNet, Number of labels=4,0002019.0611.65
TempEns SSLNumber of labels=4,0002019.0612.16
Mean Teacher SSLNumber of labels=4,0002019.0612.31
Pi Model SSLNumber of labels=4,0002019.0612.36
ImageNet-32 Fine-tunedPre-training=ImageNet-32, Number of labels=2,0002019.0612.92
SESEMI SSLBackbone=ConvNet, Number of labels=2,0002019.0614.22
SESEMI SSLBackbone=WRN, Number of labels=2,0002019.0614.45
Mean Teacher SSLNumber of labels=2,0002019.0615.73
SESEMI ASLBackbone=ConvNet, Number of labels=4,0002019.0616.15
EnAETNumber of labels=502019.1116.45
SESEMI SSLBackbone=ConvNet, Number of labels=1,0002019.0617.88
ImageNet-32 Fine-tunedPre-training=ImageNet-32, Number of labels=1,0002019.0617.96
SESEMI SSLBackbone=WRN, Number of labels=1,0002019.0618.32
Manifold MixupNumber of labels=4,0002019.0618.59
MixupNumber of labels=4,0002019.0619.67
SESEMI ASLBackbone=ConvNet, Number of labels=2,0002019.0621.53
Mean Teacher SSLNumber of labels=1,0002019.0621.55
SupervisedNumber of labels=4,0002019.0626.24
SESEMI ASLBackbone=ConvNet, Number of labels=1,0002019.0629.44
SupervisedNumber of labels=2,0002019.0633.94
Manifold MixupNumber of labels=1,0002019.0634.58