Research map / Continual learning / Online and replay-based
Bayesian Structural Adaptation for 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 Bayesian structural adaptation instead of RCL | Continual learning · SplitMNIST | test accuracy: 99.768±0.1 -> 99.819±0.1 | Helped |
| use Bayesian structural adaptation instead of RCL | Continual learning · notMNIST | test accuracy: 96.722±0.2 -> 97.152±0.2 | Helped |
| use Bayesian structural adaptation instead of RCL | Continual learning · Permuted MNIST | test accuracy: 98.005±0.2 -> 98.180±0.2 | Helped |
| use Bayesian structural adaptation instead of RCL | Continual learning · FashionMNIST | test accuracy: 98.698±0.2 -> 98.986±0.2 | Helped |
| use shared, task-masked structure | Unsupervised continual learning · MNIST | 3-KNN error: 16.6% -> 0.37% | Helped |
| use shared, task-masked structure | Unsupervised continual learning · notMNIST | 3-KNN error: 11.7% -> 0.08% | Helped |
| add a coreset (ablation) | Ours · SplitMNIST | test accuracy: 99.819±0.1 -> 99.834±0.1 | No clear effect |
| add a coreset (ablation) | Ours · notMNIST | test accuracy: 97.152±0.2 -> 97.061±0.2 | Hurt |
| add a coreset (ablation) | Ours · Permuted MNIST | test accuracy: 98.180±0.2 -> 98.163±0.3 | No clear effect |
| add a coreset (ablation) | Ours · FashionMNIST | test accuracy: 98.986±0.2 -> 98.990±0.2 | No clear effect |
| use dynamic expansion rather than truncation-based structure (ablation) | Ours · Supervised continual learning | test accuracy | Hurt |
| use one binary mask common to all tasks (ablation) | Ours · Continual learning | performance | Hurt |
Closest papers by meaning
- Learning to Continually Learn with the Bayesian Principle (ICML 2024 · similarity 0.89)
- Bayesian Adaptation of Network Depth and Width for Continual Learning (ICML 2024 · similarity 0.89)
- Temporal-Difference Variational Continual Learning (NeurIPS 2025 · similarity 0.89)
- BECAME: Bayesian Continual Learning with Adaptive Model Merging (ICML 2025 · similarity 0.88)
- A Statistical Theory of Regularization-Based Continual Learning (ICML 2024 · similarity 0.88)
- Train with Perturbation, Infer after Merging: A Two-Stage Framework for Continual Learning (NeurIPS 2025 · 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.