Research map / Continual learning / Online and replay-based
CLeAR: Continual Learning on Algorithmic Reasoning for Human-like Intelligence
Bong Gyun Kang, HyunGi Kim, Dahuin Jung, Sungroh Yoon
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 CLeAR to sequential training (ablation) | RNN · Regular CN-MA-PC-EP | CL Final accuracy (%): 47.50 -> 96.17 | Improved |
| add CLeAR to sequential training (ablation) | LSTM · Regular CN-MA-PC-EP | CL Final accuracy (%): 50.65 -> 96.49 | Improved |
| add CLeAR to sequential training (ablation) | RNN · Regular CN-MA-PC-EP | BWT: -69.80 -> -0.37 | Improved |
| add CLeAR to sequential training (ablation) | Tape-RNN · Regular EP-PC-MA-CN | CL Final accuracy (%): 47.56 -> 74.45 | Improved |
| add CLeAR to sequential training (ablation) | Stack-RNN · Regular EP-PC-MA-CN | BWT: -59.82 -> 1.45 | Improved |
| add CLeAR to sequential training (ablation) | Stack-RNN · Regular CN-MA-PC-EP | BWT: -69.04 -> -10.40 | Improved |
| change task order from CN-MA-PC-EP to EP-PC-MA-CN with CLeAR | RNN · Regular in-hierarchy | CL Final accuracy (%): 96.17 -> 99.80 | Improved |
| change task order from CN-MA-PC-EP to EP-PC-MA-CN with CLeAR | Stack-RNN · Regular in-hierarchy | CL Final accuracy (%): 87.09 -> 99.28 | Improved |
| continual learning with CLeAR instead of joint training | LSTM · High-correlation Modular Arithmetic | accuracy (%): 93.07 -> 96.25 | Improved |
| continual learning with CLeAR instead of joint training | Stack-RNN · High-correlation Bucket Sort | accuracy (%): 93.49 -> 92.44 | No clear effect |
| continual learning with CLeAR instead of joint training | Tape-RNN · In-hierarchy Regular EP-PC-MA-CN | accuracy (%): 59.61 -> 74.45 | Improved |
| continual learning with CLeAR instead of joint training | Stack-RNN · In-hierarchy Context Free DT-RS-SM-CO | accuracy (%): 69.64 -> 69.95 | No clear effect |
Closest papers by meaning
- A Comprehensive Survey of Continual Learning: Theory, Method and Application (TPAMI 2024 · similarity 0.89)
- The Ideal Continual Learner: An Agent That Never Forgets (ICML 2023 · similarity 0.88)
- A Unified and General Framework for Continual Learning (ICLR 2024 · similarity 0.87)
- Language Guided Concept Bottleneck Models for Interpretable Continual Learning (CVPR 2025 · similarity 0.87)
- Interactive Continual Learning: Fast and Slow Thinking (CVPR 2024 · similarity 0.86)
- Enhancing Visual Continual Learning with Language-Guided Supervision (CVPR 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.
