Research map / Continual learning / Online and replay-based
Plug-and-Play Compositionality for Boosting Continual Learning with Foundation Models
Weiduo Liao, Fei Han, Hisao Ishibuchi, Qingfu Zhang, Ying Wei
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 CompSLOT plug-in | ADAM+adapter · CGQA 10-10 tasks | AA: 41.930 -> 49.480 | Improved |
| add CompSLOT plug-in | CPrompt · CGQA 10-10 tasks | AA: 46.753 -> 48.537 | Improved |
| add CompSLOT plug-in | RanPAC · CGQA 10-10 tasks | AA: 65.810 -> 66.753 | Improved |
| add CompSLOT plug-in | FOSTER* · CGQA 10-10 tasks | AA: 60.863 -> 66.290 | Improved |
| add CompSLOT plug-in | EASE · CGQA 10-10 tasks | forgetting FF (lower better): 18.215 -> 22.470 | Hurt |
| add CompSLOT plug-in | DER* · CGQA 10-10 tasks | compositional Hn: 90.119 -> 88.986 | Mixed |
| add CompSLOT plug-in | CPrompt · CGQA 10-10 tasks | compositional Hn: 78.063 -> 79.091 | Improved |
| increase parameters to match CompSLOT without concept learning (ablation) | RanPAC · CGQA 10-10 tasks | AA: 65.810 -> 65.080 | No clear effect |
| increase prompt length to match CompSLOT without concept learning (ablation) | CPrompt · CGQA 10-10 tasks | AA: 46.753 -> 46.300 | No clear effect |
| add primitive loss and primitive selection (vs slot averaging) (ablation) | CompSLOT on RanPAC · CGQA 10-10 tasks | AA: 58.220 -> 66.753 | Improved |
| add primitive loss and primitive selection (vs slot averaging) (ablation) | CompSLOT on CPrompt · CGQA 10-10 tasks | AA: 40.230 -> 48.537 | Improved |
| use softmax slot weighting instead of cosine (ablation) | CompSLOT on RanPAC · CGQA 10-10 tasks | AA: 63.910 -> 66.753 | Improved |
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.
- RanPAC — RanPAC: Random Projections and Pre-trained Models for Continual Learning (NeurIPS 2023) · better on 2 of 2 shared results
- FOSTER* — FOSTER: Feature Boosting and Compression for Class-Incremental Learning (ECCV 2022) · better on 1 of 1 shared result
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
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 2024 · similarity 0.90)
- Semantic Concept Steering Breaks the Explanation Drift Loop in Continual Learning (NeurIPS 2026 · similarity 0.90)
- Language Guided Concept Bottleneck Models for Interpretable Continual Learning (CVPR 2025 · similarity 0.89)
- Unlocking Compositional Generalization in Continual Few-Shot Learning (NeurIPS 2026 · similarity 0.88)
- A continual learning survey: Defying forgetting in classification tasks (TPAMI 2021 · similarity 0.88)
- Convolutional Prompting meets Language Models for Continual Learning (CVPR 2024 · similarity 0.87)
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.
