Research map / Learning theory and algorithms / Combinatorial and search optimization
Stepping Stones to Inductive Synthesis of Low-Level Looping Programs
Christopher D. Rosin
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 Delayed Acceptance instead of Basic Hillclimbing | MAKESPEARE · Count Odds | perfect test-set generalization rate (%): 42 -> 80 | Improved |
| Use Delayed Acceptance instead of Basic Hillclimbing | MAKESPEARE · Collatz Numbers | perfect test-set generalization rate (%): 0 -> 60 | Improved |
| Use Delayed Acceptance instead of Basic Hillclimbing | MAKESPEARE · Slow Sort | perfect test-set generalization rate (%): 5 -> 60 | Improved |
| Use Delayed Acceptance instead of Basic Hillclimbing | MAKESPEARE · Prod Sq of Elem | perfect test-set generalization rate (%): 65 -> 36 | Hurt |
| Use Delayed Acceptance instead of Basic Hillclimbing | MAKESPEARE · Sum of Squares | perfect test-set generalization rate (%): 1 -> 1 | No clear effect |
| Simplify to the final program from each training-success run rather than averaging all accepted training-success programs | MAKESPEARE · 18 solved benchmarks | training successes that generalize perfectly (%): 79.9% -> 84.3% | Improved |
| Select only runs attaining minimal program size | MAKESPEARE · 18 solved benchmarks | training successes that generalize perfectly (%): 100.0% | Improved |
| Relax the Fast Sort loopcount bound from 2n lg n to n^(5/3) | MAKESPEARE · Fast Sort | perfect test-set generalization | Improved |
| Restrict the TIS-100 operand range to P=401 | MAKESPEARE · Image Test Pattern 2 | program instructions: 9 instructions | Improved |
Closest papers by meaning
- Program Synthesis with Best-First Bottom-Up Search (Abstract Reprint) (AAAI 2024 · similarity 0.86)
- Eco Search: A No-delay Best-First Search Algorithm for Program Synthesis (AAAI 2025 · similarity 0.85)
- Solver-Aware Decompositions for Programming-by-Example: When Dividing Requires Knowing how to Conquer (NeurIPS 2026 · similarity 0.84)
- Towards General Loop Invariant Generation: A Benchmark of Programs with Memory Manipulation (NeurIPS 2024 · similarity 0.84)
- Synthetic Programming Elicitation for Text-to-Code in Very Low-Resource Programming and Formal Languages (NeurIPS 2024 · similarity 0.84)
- Program Synthesis via Test-Time Transduction (NeurIPS 2025 · similarity 0.84)
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.
