Research map / Learning theory and algorithms / Combinatorial and search optimization
Differentiable Inductive Logic Programming for Structured Examples
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 multiple clause weights and softor composition instead of 2-d pair weights | Differentiable ILP · Member, Plus, Append, Delete, and Subtree | number of parameters: Member: $24$ vs $144$; Plus: $120$ vs $1600$; Append: $150$ vs $2500$; Delete: $150$ vs $2500$; Subtree: $80$ vs $400$ | Helped |
| Use multiple clause weights and softor composition instead of 2-d pair weights | Differentiable ILP · Member, Plus, Append, Delete, and Subtree | runtime [s]: Member: ${\bf 0.015}$ vs $0.12$; Plus: ${\bf 0.03 }$ vs $6.91$; Append: ${\bf 0.09 }$ vs $5.18$; Delete: ${\bf 0.06}$ vs | Helped |
| Use example-guided beam search with clause refinement rather than naive clause generation | Differentiable program searching · Append and Delete | AUC: AUC scores of $1.0$ with fewer clauses | Helped |
| Use backward-chaining enumeration from examples and generated clauses | Differentiable ILP with structured examples · Member, Plus, Append, Delete, and Subtree | number of enumerated ground atoms: Member $228$; Plus $1857$; Append $2899$; Delete $2513$; Subtree $2172$ | Helped |
| Train on examples with 10% mislabeled training data | Proposed framework · Member and Subtree | mean-squared test error: less than $0.05$ | Helped |
Closest papers by meaning
- Differentiable Learning of Logical Rules for Knowledge Base Reasoning (NeurIPS 2017 · similarity 0.90)
- End-to-End Differentiable Proving (NeurIPS 2017 · similarity 0.88)
- Differentiable Rule Induction from Raw Sequence Inputs (ICLR 2025 · similarity 0.86)
- Inductive Learning of Logical Theories with LLMs: A Expressivity-graded Analysis (AAAI 2025 · similarity 0.85)
- Learning MDL Logic Programs from Noisy Data (AAAI 2024 · similarity 0.85)
- Enhancing Reasoning Capabilities of LLMs via Principled Synthetic Logic Corpus (NeurIPS 2024 · similarity 0.85)
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.