Research map / Continual learning / Online and replay-based
Graph-Based Continual Learning
Read the paperCheck 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 learned random context and context-target graphs with graph regularization | GCL · Permuted MNIST | ACC: 79.68 -> 82.36 | Helped |
| use learned random context and context-target graphs with graph regularization | GCL · Rotated MNIST | ACC: 83.56 -> 86.37 | Helped |
| use learned random context and context-target graphs with graph regularization | GCL · Split SVHN | ACC: 45.41 -> 60.68 | Helped |
| use learned random context and context-target graphs with graph regularization | GCL · Split CIFAR10 | ACC: 29.94 -> 49.62 | Helped |
| use learned random context and context-target graphs with graph regularization | GCL · Split CIFAR100 | ACC: 69.40 -> 74.51 | Helped |
| use learned random context and context-target graphs with graph regularization | GCL · Split MiniImageNet | ACC: 58.74 -> 61.54 | Helped |
| add graph regularization to preserve previously learned context edges (ablation) | GCL · Split CIFAR10 | average accuracy: 44.04 -> 49.62 | Helped |
| add multiple graph samples and average predictions (ablation) | GCL without graph regularization · Split CIFAR10 | average accuracy: 42.08 -> 44.04 | Helped |
| use random rather than deterministic context and context-target graphs (ablation) | GCL without graph regularization and multiple graph samples · Split CIFAR10 | average accuracy: 30.50 -> 42.08 | Helped |
| avoid an extreme graph-regularization weight (lambda_G = 1000) (ablation) | GCL · Split CIFAR10 | average accuracy | Helped |
Closest papers by meaning
- Unlocking the Power of Rehearsal in Continual Learning: A Theoretical Perspective (ICML 2025 · similarity 0.89)
- Predicting the Susceptibility of Examples to Catastrophic Forgetting (ICML 2025 · similarity 0.89)
- A Unified and General Framework for Continual Learning (ICLR 2024 · similarity 0.89)
- A Topology-aware Graph Coarsening Framework for Continual Graph Learning (NeurIPS 2024 · similarity 0.89)
- Exploring Rationale Learning for Continual Graph Learning (AAAI 2025 · similarity 0.88)
- Continual Learning with Global Alignment (NeurIPS 2024 · similarity 0.88)
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.