List your company
ResearchFormerly WizWand

5-way 1-shot Classification on Omniglot (test)

99.5AccuracyiMAML
38.181654.100870.0285.9392Oct 4, 2018Jan 4, 2019Apr 7, 2019Jul 9, 2019Oct 10, 2019Jan 11, 2020Apr 13, 2020
Updated 8mo ago

Evaluation results

MethodLinks
iMAMLInner optimizer=Hessian-Free2019.0999.5——
iMAMLInner optimizer=Gradient Descent (GD)2019.0999.16——
MAML2019.0998.7——
Oracle-ProtoNetscontrol=true2018.1098.35——
ProtoNetsClustering=N/A, Supervised=true2020.0498.35——
first-order MAML2019.0998.3——
Reptile2019.0997.68——
Oracle-MAMLcontrol=true2018.1094.46——
MAMLClustering=N/A, Supervised=true2020.0494.46——
ULDA-ProtoNetsClustering=N/A2020.0491——
ULDA-MetaOptNetClustering=N/A2020.0490.51——
AAL-MAML++Clustering=N/A2020.0488.4——
AAL-ProtoNetsClustering=N/A2020.0484.66——
UMTRAClustering=N/A2020.0483.8——
ACAI CACTUS-MAMLEmbedding=ACAI, k=500, P=1002018.1068.84——
CACTUS-MAMLClustering=ACAI2020.0468.84——
ACAI CACTUS-ProtoNetsEmbedding=ACAI, k=500, P=1002018.1068.12——
CACTUS-ProtoNetsClustering=ACAI2020.0468.12——
ACAI linear classifierEmbedding=ACAI, Protocol=linear classifier2018.1061.08——
BiGAN CACTUS-MAMLEmbedding=BiGAN, k=500, P=1002018.1058.18——
ACAI knn-nearest neighborsEmbedding=ACAI, Protocol=knn-nearest neighbors2018.1057.46——
ACAI cluster matchingEmbedding=ACAI, Protocol=cluster matching, k=5002018.1054.94——
BiGAN CACTUS-ProtoNetsEmbedding=BiGAN, k=500, P=1002018.1054.74——
Training from scratch2018.1052.5——
Training from scratchClustering=N/A2020.0452.5——
ACAI MLP with dropoutEmbedding=ACAI, Protocol=MLP with dropout2018.1051.95——
BiGAN knn-nearest neighborsEmbedding=BiGAN, Protocol=knn-nearest neighbors2018.1049.55——
knn-nearest neighborsClustering=DeepCluster2020.0449.55——
BiGAN linear classifierEmbedding=BiGAN, Protocol=linear classifier2018.1048.28——
linear classifierClustering=DeepCluster2020.0448.28——
BiGAN cluster matchingEmbedding=BiGAN, Protocol=cluster matching, k=5002018.1043.96——
cluster matchingClustering=DeepCluster2020.0443.96——
BiGAN MLP with dropoutEmbedding=BiGAN, Protocol=MLP with dropout2018.1040.54——
MLP with dropoutClustering=DeepCluster2020.0440.54——
BASELINE CLASSIFIERMatching Function=Cosine, Fine-tuned=N2016.06—8095
BASELINE CLASSIFIERMatching Function=Cosine, Fine-tuned=Y2016.06—82.398.4
BASELINE CLASSIFIERMatching Function=Softmax, Fine-tuned=Y2016.06—8697.6
BASELINE CLASSIFIERMatching Function=Cosine, Fine-tuned=N2016.06—69.589.1
BASELINE CLASSIFIERMatching Function=Cosine, Fine-tuned=Y2016.06—70.692
BASELINE CLASSIFIERMatching Function=Softmax, Fine-tuned=Y2016.06—72.992.3
CONVOLUTIONAL SIAMESE NETMatching Function=Cosine, Fine-tuned=N2016.06—96.798.4
CONVOLUTIONAL SIAMESE NETMatching Function=Cosine, Fine-tuned=Y2016.06—97.398.4
CONVOLUTIONAL SIAMESE NETMatching Function=Cosine, Fine-tuned=N2016.06—8896.5
CONVOLUTIONAL SIAMESE NETMatching Function=Cosine, Fine-tuned=Y2016.06—88.197
GNN2019.02—99.2—
MAML2019.02—98.7—
MAML++2019.02—99.47—
MANN (NO CONV)Matching Function=Cosine, Fine-tuned=N2016.06—82.894.9
Matching NetworksMatching Function=Cosine, Fine-tuned=N2016.06—98.198.9
Matching NetworksMatching Function=Cosine, Fine-tuned=Y2016.06—97.998.7
Matching NetworksMatching Function=Cosine, Fine-tuned=N2016.06—93.898.5
Matching NetworksMatching Function=Cosine, Fine-tuned=Y2016.06—93.598.7
MC12019.02—99.47—
MC22019.02—99.77—
MC2Ensemble=3 model2019.02—99.97—
Meta-SGD2019.02—99.53—
PIXELSMatching Function=Cosine, Fine-tuned=N2016.06—41.763.2
PIXELSMatching Function=Cosine, Fine-tuned=N2016.06—26.742.6
SNAIL2019.02—99.07—