Research map / Continual learning / Class-incremental
Temporal Imbalance of Positive and Negative Supervision in Class-Incremental Learning
Jinge Ma, Fengqing Zhu
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add TAL to iCaRL | iCaRL · ImageNet-100 (10-task) | average accuracy: 43.71 -> 52.19 | Improved |
| add TAL to DER | DER · CIFAR-100 (10-task) | last accuracy: 50.75 -> 53.82 | Improved |
| add TAL to TagFex | TagFex · Food101 (10-task) | last accuracy: 66.67 -> 68.64 | Improved |
| add TAL to FOSTER | FOSTER · Food101 (10-task) | average accuracy: 63.88 -> 64.61 | Improved |
| add TAL to MEMO | MEMO · CIFAR-100 (10-task) | last accuracy: 51.70 -> 52.85 | Improved |
| use TAL with r=1.0, lambda=0.995 instead of CE (ablation) | ER · CIFAR-100 (10-task) | average accuracy: 59.96 -> 63.36 | Improved |
| use large steepness r=5.0 in TAL (ablation) | ER + TAL · CIFAR-100 (10-task) | average accuracy: 59.96 -> 56.70 (lambda=0.99) | Hurt |
| use small r=0.2 in TAL (ablation) | ER + TAL · CIFAR-100 (10-task) | average accuracy: 59.96 -> 61.30 | Improved |
| tune lambda from 0.995 to 0.9995 at r=1.0 (ablation) | ER + TAL · CIFAR-100 (10-task) | average accuracy: 63.36 -> 61.79 | Hurt |
| TAL's reweighting of negative supervision per class | iCaRL/DER/MEMO/TagFex · ImageNet-100 final-task class accuracy | per-class recall: old classes up, newest classes down slightly | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- FOSTER — FOSTER: Feature Boosting and Compression for Class-Incremental Learning (ECCV 2022) · better on 3 of 3 shared results
- iCaRL — iCaRL: Incremental Classifier and Representation Learning (CVPR 2017) · better on 3 of 3 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
- Gradient Reweighting: Towards Imbalanced Class-Incremental Learning (CVPR 2024 · similarity 0.91)
- Defying Imbalanced Forgetting in Class Incremental Learning (AAAI 2024 · similarity 0.90)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.88)
- Make Domain Shift a Catastrophic Forgetting Alleviator in Class-Incremental Learning (AAAI 2025 · similarity 0.87)
- Addressing Imbalanced Domain-Incremental Learning through Dual-Balance Collaborative Experts (ICML 2025 · similarity 0.87)
- Rebalancing Multi-Label Class-Incremental Learning (AAAI 2025 · similarity 0.86)
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.
