Research map / Continual learning / Class-incremental
Mnemonics Training: Multi-Class Incremental Learning Without Forgetting
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 mnemonics training with LUCIR instead of herding exemplars | LUCIR · CIFAR-100, 25 phases | average accuracy (%): 57.54 -> 60.96 | Helped |
| Use mnemonics training with LUCIR instead of herding exemplars | LUCIR · ImageNet-Subset, 25 phases | average accuracy (%): 61.44 -> 69.74 | Helped |
| Use mnemonics training with LUCIR instead of herding exemplars | LUCIR · ImageNet, 25 phases | average accuracy (%): 56.56 -> 61.00 | Helped |
| Add old exemplar adjustment (ablation) | Mnemonics training · CIFAR-100 and ImageNet-Subset, 5, 10, and 25 phases | average accuracy (%): 63.25 -> 63.34; 61.86 -> 62.28; 60.46 -> 60.96; 71.91 -> 72.58; 71.08 -> 71.37; 69.68 -> 69.74 | Helped |
| Add old exemplar adjustment (ablation) | Mnemonics training · CIFAR-100 and ImageNet-Subset, 5, 10, and 25 phases | forgetting rate (%): 9.80 -> 10.91; 13.44 -> 13.38; 16.68 -> 15.22; 18.27 -> 17.40; 18.08 -> 17.08; 20.96 -> 20.83 | Mixed |
| Add distillation loss (ablation) | LUCIR w/ mnemonics · CIFAR-100 and ImageNet-Subset, 5, 10, and 25 phases | average accuracy (%): 50.86 -> 63.34; 50.40 -> 62.28; 49.69 -> 60.96; 65.74 -> 72.58; 64.63 -> 71.37; 65.21 -> 69.74 | Helped |
| Add distillation loss (ablation) | LUCIR w/ mnemonics · CIFAR-100 and ImageNet-Subset, 5, 10, and 25 phases | forgetting rate (%): 38.70 -> 10.91; 36.94 -> 13.38; 34.38 -> 19.80; 29.32 -> 17.40; 27.44 -> 17.08; 29.64 -> 20.83 | Helped |
| Use random exemplars instead of herding exemplars (ablation) | Exemplar selection · CIFAR-100, 5 phases | average accuracy (%): 62.64 -> 62.96 | Hurt |
| Use random exemplars instead of herding exemplars (ablation) | Exemplar selection · ImageNet-Subset, 25 phases | average accuracy (%): 67.46 -> 69.61 | Hurt |
Closest papers by meaning
- Class Incremental Learning with Multi-Teacher Distillation (CVPR 2024 · similarity 0.87)
- Pre-trained Vision and Language Transformers Are Few-Shot Incremental Learners (CVPR 2024 · similarity 0.85)
- Federated Class-Incremental Learning: A Hybrid Approach Using Latent Exemplars and Data-Free Techniques to Address Local and Global Forgetting (ICLR 2025 · similarity 0.85)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.85)
- Tripartite Weight-Space Ensemble for Few-Shot Class-Incremental Learning (CVPR 2025 · similarity 0.85)
- Prospective Representation Learning for Non-Exemplar Class-Incremental Learning (NeurIPS 2024 · similarity 0.85)
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.