Research map / Continual learning / Online and replay-based
Detect, Decide, Unlearn: A Transfer-Aware Framework for Continual Learning
Yiwen Wang, Diana Benavides-Prado, Yun Sing Koh
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 DEDUCE (gradient-conflict detection) | DER++ · CIFAR-100 CIL | ACC (%): 36.8 -> 39.8 | Improved |
| add DEDUCE (gradient-conflict detection) | HAL · CIFAR-100 CIL | ACC (%): 11.6 -> 24.8 | Improved |
| add DEDUCE (gradient-conflict detection) | DER++ · Tiny-ImageNet CIL | ACC (%): 15.6 -> 22.1 | Improved |
| use gradient-conflict detection instead of transferability bound (ablation) | DEDUCE on DER++ · CIFAR-100 CIL | ACC (%): 37.8 -> 39.8 | Improved |
| add Local Unlearning Module (ablation) | DER++ · CIFAR-100 CIL | ACC (%): 36.8 -> 39.4 | Improved |
| add Global Unlearning Module (ablation) | DER++ · CIFAR-100 CIL | ACC (%): 36.8 -> 37.7 | Improved |
| add GUM only versus full DEDUCE (ablation) | DER++ · CORE-50 CIL | ACC (%): 41.2 -> 42.1 | Mixed |
| add FIM regularizer Lreg (ablation) | DEDUCE on DER++ · Tiny-ImageNet CIL | ACC (%): 19.6 -> 22.1 | Improved |
| raise local unlearning rate delta from 0.0001 to 0.01 (0.05 degrades) (ablation) | DEDUCE on DER++ · CIFAR-100 CIL | ACC (%): 37.77 -> 44.08 -> 31.04 | Mixed |
| raise global unlearning rate phi from 1e-5 to 5e-4 (ablation) | DEDUCE on DER++ · CIFAR-100 CIL | ACC (%): 39.84 -> 31.26 | Hurt |
| stricter trigger tolerance epsilon -0.2 instead of 0.0 (ablation) | DEDUCE on DER++ · CIFAR-100 CIL | ACC (%): 39.84 -> 35.97 | Hurt |
| add DEDUCE at small buffer size 100 (ablation) | ER · CIFAR-100 buffer 100 | ACC (%): 58.23 -> 62.65 | Improved |
Closest papers by meaning
- A Unified and General Framework for Continual Learning (ICLR 2024 · similarity 0.91)
- Model Sensitivity Aware Continual Learning (NeurIPS 2024 · similarity 0.89)
- BECAME: Bayesian Continual Learning with Adaptive Model Merging (ICML 2025 · similarity 0.89)
- Disentangling and mitigating the impact of task similarity for continual learning (NeurIPS 2024 · similarity 0.89)
- Language Guided Concept Bottleneck Models for Interpretable Continual Learning (CVPR 2025 · similarity 0.89)
- Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning (NeurIPS 2025 · similarity 0.89)
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.
