Research map / Continual learning / Class-incremental
Continual Prototype Evolution: Learning Online From Non-Stationary Data Streams
Matthias De Lange, Tinne Tuytelaars
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 pseudo-prototypes to the PPP-loss (ablation) | CoPE · Split-MNIST (balanced) | accuracy: 92.4 -> 93.9 | Improved |
| add pseudo-prototypes to the PPP-loss (ablation) | CoPE · Split-CIFAR10 (balanced) | accuracy: 41.3 -> 48.9 | Improved |
| add pseudo-prototypes to the PPP-loss (ablation) | CoPE · Split-CIFAR100 (balanced) | accuracy: 16.3 -> 21.6 | Improved |
| increase processing batch size from 10 to 20 (ablation) | CoPE · Split-CIFAR100 (balanced) | accuracy: 21.6 -> 21.7 | No clear effect |
| increase processing batch size from 10 to 200 (ablation) | CoPE · Split-CIFAR10 (balanced) | accuracy: 48.9 -> 37.0 | Hurt |
| increase processing batch size from 10 to 200 (ablation) | CoPE · Split-CIFAR100 (balanced) | accuracy: 21.6 -> 11.2 | Hurt |
| increase processing batch size from 10 to 200 (ablation) | CoPE · Split-MNIST (balanced) | accuracy: 93.9 -> 89.3 | Hurt |
| use prototypical PPP-loss instead of cross-entropy with same memory (ablation) | CoPE vs CoPE-CE · Split-CIFAR10 (balanced) | accuracy: 39.73 -> 48.92 | Improved |
| use prototypical PPP-loss instead of cross-entropy with same memory (ablation) | CoPE vs CoPE-CE · Split-MNIST (balanced) | accuracy: 91.77 -> 93.94 | Improved |
| use prototypical PPP-loss instead of cross-entropy with same memory (ablation) | CoPE vs CoPE-CE · Split-CIFAR100 (balanced) | accuracy: 18.33 -> 21.62 | Improved |
| use CoPE instead of reservoir replay | reservoir · Split-CIFAR10 (balanced) | accuracy: 42.48 -> 48.92 | Improved |
| use balancing memory scheme of CoPE-CE instead of reservoir (ablation) | reservoir · Split-CIFAR100 (imbalanced) | accuracy | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- iCaRL — iCaRL: Incremental Classifier and Representation Learning (CVPR 2017) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Prototype-Sample Relation Distillation: Towards Replay-Free Continual Learning (ICML 2023 · similarity 0.91)
- Online Prototype Learning for Online Continual Learning (ICCV 2023 · similarity 0.91)
- Non-exemplar Online Class-Incremental Continual Learning via Dual-Prototype Self-Augment and Refinement (AAAI 2024 · similarity 0.89)
- Long-Tail Class Incremental Learning via Independent Sub-prototype Construction (CVPR 2024 · similarity 0.87)
- Enhancing Online Continual Learning with Plug-and-Play State Space Model and Class-Conditional Mixture of Discretization (CVPR 2025 · similarity 0.86)
- Adaptive Discovering and Merging for Incremental Novel Class Discovery (AAAI 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.
