Research map / Continual learning / Class-incremental
XIL: Cross-Expanding Incremental Learning
Heayoun Choi, Hyundong Jin, Eunwoo Kim
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 domain-specialized prompts (ablation) | XEED · PACS | F-BiDoT: 45.22 -> 65.19 | Improved |
| add domain-specialized prompts (ablation) | XEED · DomainNet | F-BiDoT: 26.62 -> 33.63 | Improved |
| add synthetic image generation (ablation) | XEED · PACS | F-BiDoT: 20.91 -> 65.19 | Improved |
| add synthetic image generation (ablation) | XEED · DomainNet | F-BiDoT: 4.47 -> 33.63 | Improved |
| add synthetic image generation (ablation) | XEED · Office-31 | F-BiDoT: 33.62 -> 78.08 | Improved |
| add evolving prototype classifier instead of linear classifier (ablation) | XEED · PACS | F-BiDoT: 18.85 -> 65.19 | Improved |
| add evolving prototype classifier instead of linear classifier (ablation) | XEED · Office-31 | F-BiDoT: 35.60 -> 78.08 | Improved |
| add domain prototype matching for prompt selection (ablation) | XEED · PACS | F-BiDoT: 52.33 -> 65.19 | Improved |
| add domain prototype matching for prompt selection (ablation) | XEED · Office-31 | F-BiDoT: 77.83 -> 78.08 | No clear effect |
| use lightweight prompts instead of residual adapters (ablation) | XEED · Office-31 | F-BiDoT: 67.97 -> 78.08 | Improved |
| use lightweight prompts instead of residual adapters (ablation) | XEED · Office-31 | Final accuracy: 76.71 -> 80.72 | Improved |
| split tasks into 6 finer tasks (XEED vs CPrompt) | XEED · Office-31 | F-BiDoT: 84.40 vs 70.12 | Improved |
Closest papers by meaning
- Bring Evanescent Representations to Life in Lifelong Class Incremental Learning (CVPR 2022 · similarity 0.89)
- Advancing Cross-domain Discriminability in Continual Learning of Vision-Language Models (NeurIPS 2024 · similarity 0.88)
- Navigating Semantic Drift in Task-Agnostic Class-Incremental Learning (ICML 2025 · similarity 0.87)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.87)
- Dual Consolidation for Pre-Trained Model-Based Domain-Incremental Learning (CVPR 2025 · similarity 0.87)
- Make Domain Shift a Catastrophic Forgetting Alleviator in Class-Incremental Learning (AAAI 2025 · 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.
