Research map / Learning theory and algorithms / Combinatorial and search optimization
Neural Logic Machines
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 lifted NLM reasoning instead of MemNN | Family Tree and Graph reasoning · Family Tree and Graph benchmarks | accuracy: NLM 100% / 100%; MemNN varies by task, including 59.8% / 65.2% on HasFather and 43.9% / 67.9% on 6-Connectivity | Helped |
| Use NLM instead of MemNN | General graph 2-OutDegree reasoning · 2-OutDegree | accuracy: 100% / 100% versus N/A / N/A | Helped |
| Use NLM instead of MemNN | BlocksWorld decision-making policy · BlocksWorld (m=10/50) | task completion probability: 100% / 100% versus 0% / 0% | Helped |
| Use NLM instead of MemNN | Sorting policy · Sorting (m=10/50) | task completion probability: 100% / 100% versus 100% / 90% | Helped |
| Use NLM instead of MemNN | Path-finding policy · Path (m=10/50) | task completion probability: 100% / 100% versus 45% / 12% | Helped |
| Train the perception network jointly with NLM reasoning | CNN + NLM · AdjacentToNumber0 (m=50) | accuracy: 99.4% | Helped |
| Increase the maximum number of people in training examples to 30 | NLM IsMGUncle learner · IsMGUncle graduation | graduation ratio: 20% -> 50% | Helped |
| Train on small instances and test on larger instances | NLM · Family Tree and Graph reasoning | accuracy: 100% at m = 20 and m = 100; 100% at m = 10 and m = 50 | Helped |
Closest papers by meaning
- Neuro-Symbolic Inductive Logic Programming with Logical Neural Networks (AAAI 2022 · similarity 0.90)
- Learn to Explain Efficiently via Neural Logic Inductive Learning (ICLR 2020 · similarity 0.89)
- Inductive Learning of Logical Theories with LLMs: A Expressivity-graded Analysis (AAAI 2025 · similarity 0.87)
- LogicMP: A Neuro-symbolic Approach for Encoding First-order Logic Constraints (ICLR 2024 · similarity 0.87)
- Improving Neural Logic Machines via Failure Reflection (ICML 2024 · similarity 0.86)
- Enhancing Reasoning Capabilities of LLMs via Principled Synthetic Logic Corpus (NeurIPS 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.