Research map / Continual learning / Online and replay-based
Rapid Adaptation in Online Continual Learning: Are We Evaluating It Right?
Hasan Abed Al Kader Hammoud, Ameya Prabhu, Ser-Nam Lim, Philip H.S. Torr, Adel Bibi, Bernard Ghanem
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 FIFO sampling to the full-model/FC-Only learner (vs uniform) (ablation) | OverAdapt · CGLM online accuracy | online accuracy: 34 -> 78 | Improved |
| add FIFO sampling (vs uniform) to full-model training (ablation) | ER full-model · CGLM information retention (BWT) | accuracy: 42 -> 7 | Hurt |
| add FIFO sampling (vs uniform) to full-model training (ablation) | ER full-model · CLOC information retention (BWT) | accuracy: 18 -> 1 | Hurt |
| add FIFO sampling (vs uniform) to full-model training (ablation) | ER full-model · CLOC online accuracy | online accuracy: 14 -> 24 | Improved |
| use FC-Only training instead of full-model training under FIFO (ablation) | OverAdapt · CGLM online accuracy | online accuracy: 78 -> 82 | Improved |
| use FC-Only training instead of full-model training (ablation) | OverAdapt · CGLM/CLOC online and near-future accuracy | accuracy | Mixed |
| replace online accuracy with near-future accuracy (shift S=256 CGLM, 16384 CLOC) | FIFO-sampled OverAdapt variants · CGLM and CLOC | adaptation accuracy: about 70% drop in CGLM and 20-40% on CLOC | Hurt |
| replace online accuracy with near-future accuracy | uniform-sampling variants · CGLM and CLOC | adaptation accuracy | No clear effect |
| add mixed sampling (half FIFO, half uniform) instead of uniform (ablation) | ER (Replay Only) · CGLM near-future accuracy | near-future accuracy | No clear effect |
| add mixed sampling instead of uniform (ablation) | ER (Replay Only) · CLOC near-future accuracy | near-future accuracy | Mixed |
| change weight decay (ablation) | ER (Replay Only) · CGLM | near-future accuracy and BWT | No clear effect |
| change learning rate by an order of magnitude from 0.005 (ablation) | ER (Replay Only) · CGLM | near-future accuracy and BWT | Hurt |
Closest papers by meaning
- Forgetting, Ignorance or Myopia: Revisiting Key Challenges in Online Continual Learning (NeurIPS 2024 · similarity 0.89)
- Enhancing Online Continual Learning with Plug-and-Play State Space Model and Class-Conditional Mixture of Discretization (CVPR 2025 · similarity 0.89)
- Ferret: An Efficient Online Continual Learning Framework under Varying Memory Constraints (CVPR 2025 · similarity 0.88)
- Kalman Filter for Online Classification of Non-Stationary Data (ICLR 2024 · similarity 0.87)
- Improving Plasticity in Online Continual Learning via Collaborative Learning (CVPR 2024 · similarity 0.87)
- Adaptive Retention & Correction: Test-Time Training for Continual Learning (ICLR 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.
