Research map / Label-efficient and robust learning / Recognition and long-tail
Adaptive Neural Compilation
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 |
|---|---|---|---|
| Initialise controller with slightly non-ideal compiled weights (correct argmax, nonzero off-choices) instead of exact Dirac-delta weights | ANC controller initialisation · Access/Increment/Swap/ListK/Addition/Sort (biased) | average iterations vs. ideal compiled program | Helped |
| Gradient-based optimisation of compiled program on biased data | Access (biased) · Access (biased) | average iterations: 6 -> 4 | Helped |
| Gradient-based optimisation of compiled program on biased data | Increment (biased) · Increment (biased) | average iterations: 40 -> 16 | Helped |
| Gradient-based optimisation of compiled program on biased data | Sort (biased) · Sort (biased) | average iterations: 38 -> 18 | Helped |
| Gradient-based optimisation of compiled program on biased data | Addition (biased) · Addition (biased) | success rate: 12% | Mixed |
| Gradient-based optimisation of compiled program on biased data | Swap (biased) · Swap (biased) | success rate: 27% | Mixed |
Closest papers by meaning
- Gradient-Based Program Synthesis with Neurally Interpreted Languages (ICLR 2026 · similarity 0.88)
- Learning to Compile Programs to Neural Networks (ICML 2024 · similarity 0.88)
- Learning Differentiable Programs with Admissible Neural Heuristics (NeurIPS 2020 · similarity 0.87)
- Differentiable Programs with Neural Libraries (ICML 2017 · similarity 0.87)
- Making Neural Programming Architectures Generalize via Recursion (ICLR 2017 · similarity 0.86)
- Latent Execution for Neural Program Synthesis Beyond Domain-Specific Languages (NeurIPS 2021 · 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.