AI Paper Atlas

Research map / Continual learning / Class-incremental

Heterogeneous Forgetting Compensation for Class-Incremental Learning

Jiahua Dong, Wenqi Liang, Yang Cong, Gan Sun

ICCV 2023 · Continual learning › Class-incremental · approach: Class-balanced knowledge retention · 27 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add task-semantic aggregation (TSA) block to the baseline (ablation)HFC baseline (DyTox-style losses) · CIFAR-100 (T=5, B=0%)average accuracy: 81.6 -> 83.1Improved
add gradient-balanced forgetting compensation (GFC) loss on top of TSA (ablation)HFC baseline + TSA · CIFAR-100 (T=5, B=0%)average accuracy: 83.1 -> 84.0Improved
add gradient-balanced relation distillation (GRD) loss on top of TSA and GFC (ablation)HFC · CIFAR-100 (T=5, B=0%)average accuracy: 84.0 -> 86.4Improved
add TSA, GFC and GRD to reduce forgetting heterogeneity (ablation)HFC · CIFAR-100 (T=5, B=0%)forgetting heterogeneity (FH): 128.9 -> 103.6Improved
apply FC and RD plug-and-play losses to PODNetPODNet · CIFAR-100 (T=10, B=0%, ResNet-32)top-1 average accuracy: 54.5 -> 60.0Improved
apply FC and RD plug-and-play losses to SS-ILSS-IL · CIFAR-100 (T=10, B=0%, ResNet-32)top-1 average accuracy: 59.0 -> 64.9Improved
apply FC and RD plug-and-play losses to AFCAFC · CIFAR-100 (T=10, B=0%, ResNet-32)top-1 average accuracy: 57.1 -> 63.8Improved
apply FC and RD plug-and-play losses to FOSTERFOSTER · CIFAR-100 (T=10, B=0%, ResNet-32)top-1 average accuracy: 68.7 -> 69.9Improved
apply FC and RD plug-and-play losses to DyTox with ViT-TinyDyTox · CIFAR-100 (T=10, B=0%, ViT-Tiny)top-1 average accuracy: 73.5 -> 74.0Improved
apply FC and RD losses to AFC in first taskAFC · CIFAR-100 (T=10, B=0%, ResNet-32)top-1 accuracy at first task: 90.9 -> 88.7Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy