List your company
ResearchFormerly WizWand

Image Classification on permuted MNIST (pMNIST) (test)

97.87AccuracyHNET+R
26.567645.078863.5982.1012Jan 5, 2019Mar 12, 2020May 18, 2021Jul 25, 2022Sep 30, 2023Dec 5, 2024Feb 11, 2026
Updated 8mo ago

Evaluation results

MethodLinks
HNET+RContinual Learning Scenario=CL12019.0697.87
HNET+RContinual Learning Scenario=CL32019.0697.76
HNET+RContinual Learning Scenario=CL22019.0697.6
HNET+TIRContinual Learning Scenario=CL32019.0697.59
HNET+TIRContinual Learning Scenario=CL22019.0697.58
HNET+ENTContinual Learning Scenario=CL12019.0697.57
HNET+TIRContinual Learning Scenario=CL12019.0697.57
DGRContinual Learning Scenario=CL12019.0697.51
Fine-TuningTask Identity Protocol=TGiven, Evaluation Timing=During training, Backbone=MLP-100,1002021.0397.44
DGRContinual Learning Scenario=CL22019.0697.35
su-SNNArchitecture=Spiking Neural Network2026.0297.33
PosteriorReplay-BbBBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TGiven-During2021.0396.92
PosteriorReplay-BbBBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TGiven-Final2021.0396.84
PosteriorReplay-DiracBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TGiven-During2021.0396.73
Rhythm-SNNsArchitecture=Spiking Neural Network2026.0296.73
ASRC-SNNArchitecture=Spiking Neural Network2026.0296.62
DGRContinual Learning Scenario=CL32019.0696.38
DNC+CUWMemory slots=15, Cache size=10, Controller=Single layer GRU, Hidden vector dimension=1002019.0196.3
Dilated-RNNLayers=92019.0196.1
EWCContinual Learning Scenario=CL12019.0695.96
PosteriorReplay-DiracBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TGiven-Final2021.0395.9
AntisymmetricRNN# units=128, # params=10k2019.0295.8
PR-DiracTask Identity Protocol=TGiven, Evaluation Timing=During training, Backbone=MLP-100,1002021.0395.7
DNC+UWMemory slots=15, Controller=Single layer GRU, Hidden vector dimension=1002019.0195.6
PR-BbBTask Identity Protocol=TGiven, Evaluation Timing=During training, Backbone=MLP-100,1002021.0395.44
SIContinual Learning Scenario=CL22019.0695.33
r-LSTM Full BP2019.0195.2
brf-SNNArchitecture=Spiking Neural Network2026.0295.2
PR-DiracTask Identity Protocol=TGiven, Evaluation Timing=Final, Backbone=MLP-100,1002021.0395.05
SIContinual Learning Scenario=CL12019.0694.75
Dilated-GRULayers=92019.0194.6
KRU# units=512, # params=11k2019.0294.5
EWCContinual Learning Scenario=CL22019.0694.42
PR-BbBTask Identity Protocol=TGiven, Evaluation Timing=Final, Backbone=MLP-100,1002021.0394.35
PR-RadialTask Identity Protocol=TGiven, Evaluation Timing=During training, Backbone=MLP-100,1002021.0394.31
PR-RadialTask Identity Protocol=TGiven, Evaluation Timing=Final, Backbone=MLP-100,1002021.0394.3
SRNNArchitecture=Spiking Neural Network2026.0294.3
FC uRNN# units=512, # params=270k2019.0294.1
DNCMemory slots=Best of {15, 30, 60}, Controller=Single layer GRU, Hidden vector dimension=1002019.0194
PR-SSGETask Identity Protocol=TGiven, Evaluation Timing=During training, Backbone=MLP-100,1002021.0393.58
AntisymmetricRNN w/ gating# units=128, # params=10k2019.0293.1
PR-SSGETask Identity Protocol=TGiven, Evaluation Timing=Final, Backbone=MLP-100,1002021.0392.88
HNET+ENTContinual Learning Scenario=CL22019.0692.8
LSTM# units=128, # params=68k2019.0292.6
FC uRNN# units=116, # params=16k2019.0292.1
HNET+ENTContinual Learning Scenario=CL32019.0691.75
Soft orthogonal# units=128, # params=18k2019.0291.4
uRNN2019.0191.4
GLIFArchitecture=Spiking Neural Network2026.0290.47
PR-BbBTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Ent, Backbone=MLP-100,1002021.0389.9
PR-BbBTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Conf, Backbone=MLP-100,1002021.0388.38
PosteriorReplay-BbBBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TInfer-Final (Ent)2021.0385.84
PosteriorReplay-BbBBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TInfer-Final (Conf)2021.0384.35
IRNN2019.0182
PR-RadialTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Ent, Backbone=MLP-100,1002021.0381.78
PosteriorReplay-BbBBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TInfer-Final (Agree)2021.0381.33
PR-RadialTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Conf, Backbone=MLP-100,1002021.0379.87
PR-SSGETask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Ent, Backbone=MLP-100,1002021.0378.94
PR-SSGETask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Conf, Backbone=MLP-100,1002021.0377.43
PR-RadialTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Agree, Backbone=MLP-100,1002021.0376.73
PR-DiracTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Ent, Backbone=MLP-100,1002021.0375.84
PR-DiracTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Conf, Backbone=MLP-100,1002021.0375.14
PR-BbBTask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Agree, Backbone=MLP-100,1002021.0370.52
PosteriorReplay-DiracBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TInfer-Final (Ent)2021.0370.08
PosteriorReplay-DiracBackbone=MLP-1000,1000, Stochastic regularization (SR)=true, Evaluation Protocol=TInfer-Final (Conf)2021.0369.82
PR-SSGETask Identity Protocol=TInfer, Evaluation Timing=Final, Inference Criterion=Agree, Backbone=MLP-100,1002021.0368.93
Fine-TuningTask Identity Protocol=TGiven, Evaluation Timing=Final, Backbone=MLP-100,1002021.0347.89
EWCContinual Learning Scenario=CL32019.0633.88
SIContinual Learning Scenario=CL32019.0629.31