Research map / Continual learning / Online and replay-based
CLEO: Continual Learning of Evolving Ontologies
Shishir Muralidhara, Saqib Bukhari, Georg Dr. Schneider, Didier Stricker, René Schuster
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 |
|---|---|---|---|
| extend MiB distillation to group logits of evolving parent classes with their split children (MoOn) | MiB-style distillation · Cityscapes CS-Ex1 | mIoU (All): 9.11 -> 39.31 | Improved |
| use MoOn instead of MiB | continual segmentation · PASCAL VOC VOC-Ex1 | mIoU (All): 47.05 -> 62.52 | Improved |
| use MoOn instead of MiB | continual segmentation · Mapillary Vistas MV-Ex1 | mIoU (All): 21.27 -> 21.44 | No clear effect |
| use MoOn instead of MiB | continual segmentation · Mapillary Vistas MV-Ex2 | mIoU (All): 6.45 -> 6.49 | No clear effect |
| use MoOn instead of PLOP for retained classes | continual segmentation · Cityscapes CS-Ex1 | mIoU (Retained): 13.02 -> 37.00 | Improved |
| lengthen task sequence (MV-Ex1 two tasks to MV-Ex2 eleven tasks) (ablation) | MoOn · Mapillary Vistas | mIoU (All): 21.44 -> 6.49 | Hurt |
| apply RCIL to evolving class splits | RCIL · Cityscapes CS-Ex1 | mIoU (Split): 0.00 on split classes | Hurt |
| naive fine-tuning without forgetting prevention | baseline · Cityscapes CS-Ex1 | mIoU (All): 55.62 -> 00.00 vs joint training | Hurt |
Closest papers by meaning
- CLDyB: Towards Dynamic Benchmarking for Continual Learning with Pre-trained Models (ICLR 2025 · similarity 0.86)
- Continual Learning Bench: Evaluating Frontier AI Systems in Real-World Stateful Environments (NeurIPS 2026 · similarity 0.85)
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 2024 · similarity 0.84)
- K-DeCore: Facilitating Knowledge Transfer in Continual Structured Knowledge Reasoning via Knowledge Decoupling (NeurIPS 2025 · similarity 0.84)
- Continual Relation Extraction via Sequential Multi-Task Learning (AAAI 2024 · similarity 0.84)
- CL-LoRA: Continual Low-Rank Adaptation for Rehearsal-Free Class-Incremental Learning (CVPR 2025 · similarity 0.84)
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.
