Research map / Learning theory and algorithms / Combinatorial and search optimization
Learning Randomized Reductions
Ferhat Erata, Orr Paradise, Thanos Typaldos, Timos Antonopoulos, ThanhVu Nguyen, Shafi Goldwasser, Ruzica Piskac
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 |
|---|---|---|---|
| use linear regression backend instead of PySR | V-Bitween · RSR-Bench | RSR coverage (%): 38% -> 54% | Improved |
| use linear regression backend instead of GPLearn | V-Bitween · RSR-Bench | RSR coverage (%): 32% -> 54% | Improved |
| use linear regression backend instead of MILP | V-Bitween · RSR-Bench | RSR coverage (%): 51% -> 54% | Improved |
| use linear regression backend instead of PySR | V-Bitween · RSR-Bench | average runtime (sec): 335 -> 5 | Improved |
| give LLM agent V-Bitween inference and verification tools (A-Bitween vs N-Research) | Claude-Opus-4.1 · RSR-Bench | RSR coverage (%): 64% -> 80% | Improved |
| give LLM agent V-Bitween tools | Claude-Sonnet-4 · RSR-Bench | RSR coverage (%): 60% -> 66% | Improved |
| give LLM agent V-Bitween tools | GPT-OSS-120B · RSR-Bench | RSR coverage (%): 62% -> 59%; RSR count 170 -> 157 | Mixed |
| give LLM agent V-Bitween tools | Claude-Sonnet-4 · RSR-Bench | RSR count: 191 -> 293 | Improved |
| give LLM agent V-Bitween tools | Claude-Opus-4.1 · RSR-Bench | unverified properties: 172 -> 26 | Improved |
| allow novel query functions via LLM agent (A-Bitween Opus vs V-Bitween LR) | Bitween · RSR-Bench | RSR coverage (%): 54% -> 80% | Improved |
| add LLM agent tools overhead | A-Bitween · RSR-Bench | runtime and tokens: 1.2--1.8x runtime; 2.8--3.5x tokens | Hurt |
| use A-Bitween on non-scalar algebraic structures | A-Bitween (Claude-Opus-4.1) · 17 non-scalar benchmarks | functions with verified RSR: 11 / 17 -> 17 | Improved |
Closest papers by meaning
- Auto-Regressive Next-Token Predictors are Universal Learners (ICML 2024 · similarity 0.84)
- SMILE: Bridging Continuous Optimization and Discrete Symbolic Recovery (NeurIPS 2026 · similarity 0.84)
- On the Complexity of Learning Sparse Functions with Statistical and Gradient Queries (NeurIPS 2024 · similarity 0.83)
- The Gradient of Algebraic Model Counting (AAAI 2025 · similarity 0.83)
- Tokenized Bandit for LLM Decoding and Alignment (ICML 2025 · similarity 0.83)
- Finding Simple Proofs for First-Order Optimization (NeurIPS 2026 · 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.
