Research map / Continual learning / Online and replay-based
On Learning the Geodesic Path for Incremental Learning
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 |
|---|---|---|---|
| Replace LUCIR feature distillation with GeoDL | LUCIR · CIFAR-100, 5 tasks | average accuracy (%): 63.17 -> 65.14 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · CIFAR-100, 10 tasks | average accuracy (%): 60.14 -> 65.03 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · CIFAR-100, 25 tasks | average accuracy (%): 57.54 -> 63.12 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-subset, 5 tasks | average accuracy (%): 70.84 -> 73.87 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-subset, 10 tasks | average accuracy (%): 68.32 -> 73.55 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-subset, 25 tasks | average accuracy (%): 61.44 -> 71.72 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-full, 5 tasks | average accuracy (%): 64.45 -> 65.23 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-full, 10 tasks | average accuracy (%): 61.57 -> 64.46 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · ImageNet-full, 25 tasks | average accuracy (%): 56.56 -> 62.20 | Helped |
| Replace LUCIR feature distillation with GeoDL | LUCIR · CIFAR-100, 10 tasks | forgetting rate (%): 21.34 -> 9.10 | Helped |
| Use GeoDL instead of no distillation (ablation) | CIFAR-100, 10-task classifier experiment · CIFAR-100, 10 tasks, memory 20/class | average accuracy (%): 60.46 -> 62.77 | Helped |
| Use AME rather than CNN with GeoDL (ablation) | GeoDL classifier experiment · CIFAR-100, 10 tasks, memory 20/class | average accuracy (%): 58.54 -> 62.77 | Helped |
Closest papers by meaning
- Neural Collapse Inspired Knowledge Distillation (AAAI 2025 · similarity 0.88)
- Rethinking Momentum Knowledge Distillation in Online Continual Learning (ICML 2024 · similarity 0.86)
- Fine-Grained Knowledge Selection and Restoration for Non-exemplar Class Incremental Learning (AAAI 2024 · similarity 0.86)
- MIND: Multi-Task Incremental Network Distillation (AAAI 2024 · similarity 0.86)
- Progressively Knowledge Distillation via Re-parameterizing Diffusion Reverse Process (AAAI 2024 · similarity 0.86)
- Class Incremental Learning with Multi-Teacher Distillation (CVPR 2024 · similarity 0.86)
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.