Research map / Learning theory and algorithms / Combinatorial and search optimization
Towards Synthesizing Complex Programs From Input-Output 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 |
|---|---|---|---|
| Train with curriculum learning (lessons from short to long inputs) for the neural parsing program (ablation) | Ours (LL machine parser) · While-Lang and Lambda-Lang | test accuracy | Helped |
| Use baseline end-to-end seq2seq/seq2tree/LSTM models instead of the LL machine | Neural parser on long inputs · While-Lang and Lambda-Lang Test-500/Test-5000 | test accuracy | Hurt |
| Seq2tree decoder (recursive decoding) vs seq2seq on same-length test data trained on Std-50 | End-to-end baselines · While-Lang Test-50 | accuracy: +13 points over seq2seq | Helped |
| RobustFill-style DSL program synthesis (projected accuracy) instead of the LL-machine parser | Neural program synthesis baseline · While-Lang and Lambda-Lang | test accuracy | Hurt |
| Stack/Queue/DeQue LSTM with unbounded memory baselines | End-to-end baselines · While-Lang and Lambda-Lang Std-50 | training accuracy | Hurt |
Closest papers by meaning
- Leveraging Grammar and Reinforcement Learning for Neural Program Synthesis (ICLR 2018 · similarity 0.91)
- Gradient-Based Program Synthesis with Neurally Interpreted Languages (ICLR 2026 · similarity 0.89)
- Program Synthesis via Test-Time Transduction (NeurIPS 2025 · similarity 0.88)
- HYSYNTH: Context-Free LLM Approximation for Guiding Program Synthesis (NeurIPS 2024 · similarity 0.88)
- $\mathcal{B}$-Coder: Value-Based Deep Reinforcement Learning for Program Synthesis (ICLR 2024 · similarity 0.87)
- Reinforcement Learning and Data-Generation for Syntax-Guided Synthesis (AAAI 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.