Research map / Continual learning / Online and replay-based
Information Bottleneck Based Data Correction in Continual Learning
Shuai Chen, mingyi zhang, Junge Zhang, Kaiqi Huang
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 IBCL to ER | ER · Split CIFAR-100 (buffer 200) | average accuracy: 14.35 -> 23.98 | Improved |
| add IBCL to X-DER | X-DER · Split CIFAR-10 (buffer 500) | average accuracy: 68.10 -> 74.14 | Improved |
| add IBCL to DER++ | DER++ · Split ImageNet-100 (buffer 2000) | average accuracy: 58.05 -> 59.92 | Improved |
| add IBDS (unsampled data surrogate) to ER-ACE (ablation) | ER-ACE · Split CIFAR-100 (buffer 200) | average accuracy: 26.28 -> 30.44 | Improved |
| add IBTA (task-agnostic constraint) to ER-ACE (ablation) | ER-ACE · Split CIFAR-100 (buffer 200) | average accuracy: 26.28 -> 27.36 | Improved |
| add IBTA on top of IBDS to form full IBCL (ablation) | DER++ · Split CIFAR-100 (buffer 200) | average accuracy: 31.53 -> 33.82 | Improved |
| add IBCL versus adding DualHSIC to DER++ | DER++ · Split CIFAR-10 (buffer 100) | average accuracy: 64.98 -> 66.41 | Improved |
| add sSGD regularization to ER-ACE | ER-ACE · Split miniImageNet (buffer 5000) | average accuracy: 27.92 -> 24.12 | Hurt |
| add IBCL to fine-tuning with no buffer | Fine-tuning · CKA analysis | feature similarity | Improved |
Closest papers by meaning
- A Unified and General Framework for Continual Learning (ICLR 2024 · similarity 0.90)
- Mitigating Catastrophic Forgetting in Online Continual Learning by Modeling Previous Task Interrelations via Pareto Optimization (ICML 2024 · similarity 0.89)
- Model Sensitivity Aware Continual Learning (NeurIPS 2024 · similarity 0.89)
- Summarizing Stream Data for Memory-Constrained Online Continual Learning (AAAI 2024 · similarity 0.88)
- Language Guided Concept Bottleneck Models for Interpretable Continual Learning (CVPR 2025 · similarity 0.88)
- STAR: Stability-Inducing Weight Perturbation for Continual Learning (ICLR 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.
