Research map / Learning theory and algorithms / Kernels and regression theory
Saving for the Future: Enhancing Generalization via Partial Logic Regularization
Zhaorui Tan, Yijie Hu, Xi Yang, Qiufeng Wang, Anh Nguyen, Kaizhu 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 PL-Reg instead of L-Reg | PIM · GCD (avg of 6 datasets) | unknown-class accuracy: 62.7 -> 63.7 | Improved |
| add PL-Reg to PIM | PIM · GCD (avg of 6 datasets) | all-class accuracy: 67.4 -> 69.3 | Improved |
| add PL-Reg instead of L-Reg | PIM · GCD (avg of 6 datasets) | known-class accuracy: 79.0 -> 78.8 | Hurt |
| add PL-Reg instead of L-Reg | PIM · CUB (GCD) | all-class accuracy: 65.3 -> 67.1 | Improved |
| add PL-Reg instead of L-Reg | PIM · CIFAR100 (GCD) | unknown-class accuracy: 73.2 -> 74.9 | Improved |
| add PL-Reg instead of L-Reg | PIM · Herbarium19 (GCD) | all-class accuracy: 43.7 -> 43.6 | No clear effect |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD (avg of 5 datasets) | all-class accuracy: 51.94 -> 55.10 | Improved |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD (avg of 5 datasets) | unknown-class accuracy: 27.68 -> 35.56 | Improved |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD (avg of 5 datasets) | known-class accuracy: 69.87 -> 68.07 | Hurt |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD PACS | unknown-class accuracy: 31.33 -> 42.69 | Improved |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD VLCS Caltech101 | all-class accuracy: 84.54 -> 79.95 | Hurt |
| add PL-Reg instead of L-Reg | GMDG · mDG+GCD VLCS SUN09 | unknown-class accuracy: 9.84 -> 37.42 | Improved |
Closest papers by meaning
- Generalization via Derandomization (ICML 2020 · similarity 0.84)
- LogicMP: A Neuro-symbolic Approach for Encoding First-order Logic Constraints (ICLR 2024 · similarity 0.81)
- Enhancing Reasoning Capabilities of LLMs via Principled Synthetic Logic Corpus (NeurIPS 2024 · similarity 0.80)
- An Evaluation of Approaches to Train Embeddings for Logical Inference (Student Abstract) (AAAI 2025 · similarity 0.80)
- Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning (NeurIPS 2025 · similarity 0.80)
- Skip a Layer or Loop It? Learning Program-of-Layers in LLMs (ICML 2026 · similarity 0.79)
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.
