Research map / Large language models / Agents, code and math
Efficient Rule Induction by Ignoring Pointless Rules
Andrew Cropper, David M. Cerna
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 |
|---|---|---|---|
| ignore pointless (reducible and indiscriminate) rules | POPPER · 449 ILP tasks | learning time: mean decrease 28 ± 4 minutes on 96/449 tasks | Improved |
| ignore pointless rules | POPPER · 449 ILP tasks | test accuracy: significant increase on 6/449 tasks, decrease on 5/449 | No clear effect |
| prune only reducible rules (ablation) | REDUCER · 449 ILP tasks | learning time: decrease 26 ± 5 minutes on 93/449 tasks | Improved |
| prune only indiscriminate rules (ablation) | REDUCER · 449 ILP tasks | learning time: decrease 28 ± 4 minutes on 101/449 tasks | Improved |
| add pointless-rule pruning to POPPER | POPPER · IGGP eight puzzle | time to prove optimality: 60 minutes timeout -> 12 ± 0 seconds | Improved |
| add pointless-rule pruning to POPPER | POPPER · Trains | balanced accuracy: 87 ± 9 -> 94 ± 7 | Improved |
| add pointless-rule pruning to POPPER | POPPER · IGGP | balanced accuracy: 85 ± 0 -> 85 ± 0 | No clear effect |
| overhead of checking for pointless rules | REDUCER · 449 ILP tasks | overhead: mean 2% | No clear effect |
Closest papers by meaning
- Phenomenal Yet Puzzling: Testing Inductive Reasoning Capabilities of Language Models with Hypothesis Refinement (ICLR 2024 · similarity 0.84)
- Inductive Learning of Logical Theories with LLMs: A Expressivity-graded Analysis (AAAI 2025 · similarity 0.83)
- Differentiable Rule Induction from Raw Sequence Inputs (ICLR 2025 · similarity 0.83)
- Learning MDL Logic Programs from Noisy Data (AAAI 2024 · similarity 0.83)
- LogicReward: Incentivizing LLM Reasoning via Step-Wise Logical Supervision (ICLR 2026 · similarity 0.83)
- Hypothesis Search: Inductive Reasoning with Language Models (ICLR 2024 · similarity 0.83)
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.
