Research map / Continual learning / Class-incremental
Class-Incremental Learning by Knowledge Distillation With Adaptive Feature Consolidation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use adaptive feature importance weights for distillation instead of uniform weights (ablation) | AFC · CIFAR100, 50 stages | average incremental accuracy (%): 42.21 -> 62.58 (NME); 40.81 -> 62.18 (CNN) | Helped |
| Include current-task data in the discrepancy objective along with exemplars (ablation) | AFC · CIFAR100, 50 stages, memory 5/class | average incremental accuracy (%): 40.37 -> 44.66 | Helped |
| Include current-task data in the discrepancy objective along with exemplars (ablation) | AFC · CIFAR100, 50 stages, memory 50/class | average incremental accuracy (%): 65.10 -> 65.07 | Hurt |
| Vary lambda_disc from 4.0 (ablation) | AFC (NME) · CIFAR100, 50 stages | average incremental accuracy (%): 62.58 at 4.0; 60.16 at 2.0; 62.27 at 3.0; 62.37 at 5.0; 61.83 at 6.0 | Mixed |
| Use adaptive feature importance weights for distillation instead of uniform weights (ablation) | AFC · CIFAR100, 50 stages | average incremental accuracy (%): 40.81 -> 62.18 (CNN); 42.21 -> 62.58 (NME) | Helped |
Closest papers by meaning
- Class Incremental Learning with Multi-Teacher Distillation (CVPR 2024 · similarity 0.89)
- eTag: Class-Incremental Learning via Embedding Distillation and Task-Oriented Generation (AAAI 2024 · similarity 0.89)
- Compress to One Point: Neural Collapse for Pre-Trained Model-Based Class-Incremental Learning (AAAI 2025 · similarity 0.89)
- Multi-layer Rehearsal Feature Augmentation for Class-Incremental Learning (ICML 2024 · similarity 0.88)
- Gradient Reweighting: Towards Imbalanced Class-Incremental Learning (CVPR 2024 · similarity 0.88)
- M2SD:Multiple Mixing Self-Distillation for Few-Shot Class-Incremental Learning (AAAI 2024 · similarity 0.88)
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.