List your company
ResearchFormerly WizWand

Few-shot classification on Mini-ImageNet 1-shot 5-way (test)

85.59AccuracyPTMAP-SF-SOTp
25.2740.9356.5972.25Mar 8, 2018Nov 11, 2018Jul 17, 2019Mar 22, 2020Nov 25, 2020Jul 31, 2021Apr 6, 2022
Updated 8mo ago

Evaluation results

MethodLinks
PTMAP-SF-SOTptransductive=true, backbone=wrn-28-102022.0485.59
PTMAP-SFtransductive=true, backbone=wrn-28-102022.0484.81
PTMAP-SOTttransductive=true, backbone=wrn-28-102022.0484.18
Sill-Net-SOTptransductive=true, backbone=wrn-28-102022.0483.35
PTMAP-SOTptransductive=true, backbone=wrn-28-102022.0483.19
Sill-Nettransductive=true, backbone=wrn-28-102022.0482.99
PTMAPtransductive=true, backbone=wrn-28-102022.0482.92
PTMAP-SOFTMAXtransductive=true, backbone=wrn-28-102022.0480.08
PTMAP-SOTptransductive=true, backbone=resnet-122022.0478.35
PTMAP-SOTttransductive=true, backbone=resnet-122022.0477.3
PTMAPtransductive=true, backbone=resnet-122022.0476.9
PTMAP-COSINEtransductive=true, backbone=wrn-28-102022.0474.6
ProtoNet-SOTttransductive=true, backbone=resnet-122022.0467.9
RENettransductive=false, backbone=resnet-122022.0467.6
ProtoNet-SOTptransductive=true, backbone=resnet-122022.0467.34
FEATSetup=Embedding adaptation, Backbone=ResNet2018.1266.78
FEATtransductive=false, backbone=resnet-122022.0466.78
DeepEMDtransductive=false, backbone=resnet-122022.0465.91
GCNSetup=Embedding adaptation, Backbone=ResNet2018.1264.5
DEEPSETSSetup=Embedding adaptation, Backbone=ResNet2018.1264.14
CTMBackbone=ResNet2018.1264.12
BILSTMSetup=Embedding adaptation, Backbone=ResNet2018.1263.9
SimpleShotBackbone=ResNet2018.1262.85
MetaOptNetBackbone=ResNet2018.1262.64
ProtoNettransductive=false, backbone=wrn-28-102022.0462.6
ProtoNetSetup=Instance embedding, Backbone=ResNet2018.1262.39
ProtoNettransductive=false, backbone=resnet-122022.0462.39
PFABackbone=ResNet2018.1259.6
TADAMBackbone=ResNet2018.1258.5
FEAT‡Setup=Transductive, Backbone=ConvNet, Prediction Strategy=via Semi-ProtoNet2018.1257.04
TEAMSetup=Transductive, Backbone=ConvNet2018.1256.57
FEAT†Setup=Transductive, Backbone=ConvNet, Prediction Strategy=via ProtoNet2018.1256.49
sparse-ReLU-MAMLBackbone=ResNet-12, Number of seeds=32021.1056.39
TPNSetup=Transductive, Backbone=ConvNet2018.1255.51
Semi-ProtoNet (Ours)Setup=Transductive, Backbone=ConvNet2018.1255.5
ANILBackbone=ResNet-12, Number of seeds=32021.1055.25
FEATSetup=Embedding adaptation, Backbone=ConvNet2018.1255.15
FEATSetup=Standard, Backbone=ConvNet2018.1255.15
FEATtransductive=false, backbone=conv-42022.0455.15
sparse-MAMLBackbone=ResNet-12, Number of seeds=32021.1055.02
PFABackbone=ConvNet2018.1254.53
DEEPSETSSetup=Embedding adaptation, Backbone=ConvNet2018.1254.41
ProtoNet-SOTptransductive=true, backbone=conv-42022.0454.01
MAMLBackbone=ResNet-12, Number of seeds=32021.1053.91
ProtoNet-SOTttransductive=true, backbone=conv-42022.0453.7
GCNSetup=Embedding adaptation, Backbone=ConvNet2018.1253.25
ProtoNetSetup=Instance embedding, Backbone=ConvNet2018.1252.61
ProtoNetSetup=Standard, Backbone=ConvNet2018.1252.61
MAML++Inner Steps=52018.1052.15
MAML++Training Mode=Supervised Upper Bound2019.0252.15
BILSTMSetup=Embedding adaptation, Backbone=ConvNet2018.1252.13
MAML++Inner Steps=42018.1051.65
MAML++Inner Steps=22018.1051.49
RelationNetBackbone=ConvNet2018.1251.38
MetaOptNetBackbone=ResNet-12, Number of seeds=3, Regularization=None2021.1051.13
MAML++Inner Steps=32018.1051.11
MAML++Inner Steps=12018.1051.05
Meta-SGDInner Steps=12018.1050.47
Semi-ProtoNetSetup=Transductive, Backbone=ConvNet2018.1250.41
ProtoNetsTraining Mode=Supervised Upper Bound2019.0250.16
ReptileTransduction=true2018.0349.97
SimpleShotBackbone=ConvNet2018.1249.69
ProtoNetBackbone=ConvNet2018.1249.42
RelationNettransductive=false, backbone=conv-42022.0449.31
Meta-Networks2018.1049.21
ProtoNettransductive=false, backbone=conv-42022.0449.1
MAMLTransduction=true2018.0348.7
MAML (original paper)Inner Steps=52018.1048.7
MAMLBackbone=ConvNet2018.1248.7
MAML (local reproduction)Inner Steps=52018.1048.25
1st-order MAMLTransduction=true2018.0348.07
ReptileTransduction=false2018.0347.07
MAMLTraining Mode=Supervised Upper Bound2019.0246.81
MAMLtransductive=false, backbone=conv-42022.0446.47
Matching Nets2018.1043.56
MatchNetBackbone=ConvNet2018.1243.4
UMTRA + AutoAugmentLearning Paradigm=Unsupervised, Augmentation=AutoAugment2019.0239.93
DeepCluster CACTUS-MAMLLearning Paradigm=Unsupervised, Clusterer=DeepCluster2019.0239.9
DeepCluster CACTUS-ProtoNetsLearning Paradigm=Unsupervised, Clusterer=DeepCluster2019.0239.18
AAL-ProtoNetsLearning Paradigm=Unsupervised, Algorithm=ProtoNets2019.0237.67
AAL-MAML++Learning Paradigm=Unsupervised, Algorithm=MAML++2019.0233.3
Training from scratchLearning Paradigm=Unsupervised2019.0227.59