Research map / Continual learning / Class-incremental
Not Just Selection, but Exploration: Online Class-Incremental Continual Learning via Dual View Consistency
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 Dual View Consistency with maximized mutual information between two transformed views (ablation) | ER on CIFAR-100 · CIFAR-100, M=1k | Average Accuracy: 11.6 ± 0.5 -> 18.7 ± 0.8 | Helped |
| add Maximally Gradient Interfered retrieval (ablation) | ER + DVC on CIFAR-100 · CIFAR-100, M=1k | Average Accuracy: 18.7 ± 0.8 -> 19.7 ± 0.7 | Helped |
| add Maximally Gradient Interfered retrieval (ablation) | ER + DVC on CIFAR-100 · CIFAR-100, M=5k | Average Accuracy: 22.4 ± 1.4 -> 24.1 ± 0.8 | Helped |
| add Dual View Consistency (ablation) | ER on CIFAR-100 · CIFAR-100, M=1k | Average Forgetting: 39.1 ± 0.9 -> 31.3 ± 1.0 | Helped |
| add Maximally Gradient Interfered retrieval (ablation) | ER + DVC on CIFAR-100 · CIFAR-100, M=5k | Average Forgetting: 29.2 ± 1.0 -> 26.1 ± 0.5 | Helped |
| send the two augmented views to the network separately (ablation) | ER on Mini-ImageNet · Mini-ImageNet, M=1k | Average Accuracy: 10.2 ± 0.5 -> 11.7 ± 0.5 | Helped |
| send the two augmented views to the network separately (ablation) | ER on Mini-ImageNet · Mini-ImageNet, M=5k | Average Accuracy: 16.3 ± 0.9 -> 12.7 ± 1.4 | Hurt |
| combine the two augmented views into one training batch (ablation) | ER on Mini-ImageNet · Mini-ImageNet, M=1k | Average Accuracy: 10.2 ± 0.5 -> 12.3 ± 0.8 | Helped |
| combine the two augmented views into one training batch (ablation) | ER on Mini-ImageNet · Mini-ImageNet, M=2k | Average Accuracy: 12.9 ± 0.8 -> 13.3 ± 1.0 | No clear effect |
| combine the two augmented views into one training batch (ablation) | ER on Mini-ImageNet · Mini-ImageNet, M=5k | Average Accuracy: 16.3 ± 0.9 -> 14.8 ± 0.9 | Hurt |
| use DVC with MGI rather than the strongest replay baseline | online class-incremental learning · CIFAR-10, M=0.2k | Average Accuracy: 29.6 ± 1.0 -> 45.4 ± 1.4 | Helped |
| use DVC with MGI rather than the strongest replay baseline | online class-incremental learning · Mini-ImageNet, M=5k | Average Accuracy: 19.5 ± 0.5 -> 19.1 ± 0.9 | Hurt |
Closest papers by meaning
- Non-exemplar Online Class-Incremental Continual Learning via Dual-Prototype Self-Augment and Refinement (AAAI 2024 · similarity 0.91)
- Online Class-Incremental Continual Learning with Adversarial Shapley Value (AAAI 2021 · similarity 0.90)
- Revisiting Generative Replay for Class Incremental Object Detection (CVPR 2025 · similarity 0.88)
- AnaCP: Toward Upper-Bound Continual Learning via Analytic Contrastive Projection (NeurIPS 2025 · similarity 0.88)
- Enhancing Online Continual Learning with Plug-and-Play State Space Model and Class-Conditional Mixture of Discretization (CVPR 2025 · similarity 0.87)
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 2024 · similarity 0.87)
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.