List your company
ResearchFormerly WizWand

Class Incremental Learning on CIFAR100 10 steps B0 (test)

80.41Avg Top-1 AccBound
57.134863.177469.2275.2626Dec 29, 2022Jul 30, 2023Feb 29, 2024Sep 29, 2024May 1, 2025Nov 30, 2025Jul 2, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Bound#Paras=11.2, Rehearsal Setting=true2022.1280.41———
TCIL#Paras=64.1, Rehearsal Setting=true2022.1277.3—66.41—
DRDNPar.=10.772026.0777.19—65.4-11.2
TCIL-Lite#Paras=14.5, Rehearsal Setting=true2022.1276.74—66.66—
DKTPar.=11.032026.0775.83—63.04-13.8
DER#Paras=61.6, Rehearsal Setting=true2022.1275.36—65.22—
DERPar.=112.272026.0775.36—65.22-12.1
DyTox#Paras=10.7, Rehearsal Setting=true2022.1273.66—60.67—
DyToxPar.=10.732026.0773.66—60.67-14.7
WA#Paras=11.2, Rehearsal Setting=true2022.1269.46—53.78—
WAPar.=11.222026.0769.46—53.78-18.2
BiC#Paras=11.2, Rehearsal Setting=true2022.1268.8—53.54—
BiCPar.=11.222026.0768.8—53.54-17.7
RPSNet#Paras=56.5, Rehearsal Setting=true2022.1268.6—57.05—
iCaRL#Paras=11.2, Rehearsal Setting=true2022.1265.27—50.74—
iCaRLPar.=11.222026.0765.27—50.74-21.3
UCIR#Paras=11.2, Rehearsal Setting=true2022.1258.66—43.39—
UCIRPar.=11.222026.0758.66—43.39-22.1
PODNet#Paras=11.2, Rehearsal Setting=true2022.1258.03—41.05—
PODNetPar.=11.222026.0758.03—41.05-22.6
BiCBackbone=modified 32-layer ResNet, #Paras=0.46M2021.03—66.15——
BiC2022.04—65.08——
BoundPar.=10.722026.07——81.49—
COIL2022.04—65.48——
DERBackbone=modified 32-layer ResNet, #Paras=0.52M2021.03—69.41——
DERBackbone models=Same number as incremental sessions2022.04—69.74——
DER (without pruning)Backbone=modified 32-layer ResNet, #Paras=2.53M2021.03—71.29——
FOSTERBackbone models=Single backbone2022.04—72.9——
iCaRLBackbone=modified 32-layer ResNet, #Paras=0.46M2021.03—64.04——
iCaRL2022.04—64.42——
Joint Training (Bound)2022.04—80.4——
PODNet2022.04—55.22——
WABackbone=modified 32-layer ResNet, #Paras=0.46M2021.03—67.25——
WA2022.04—67.08——