Research map / Learning theory and algorithms / Combinatorial and search optimization
Deep symbolic regression: Recovering mathematical expressions from data via risk-seeking policy gradients
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 DSR rather than the strongest baseline on the Nguyen suite | Symbolic regression · Nguyen suite average | recovery rate: 83.6% vs 75.2% | Helped |
| Replace the risk-seeking objective with standard policy gradients (ablation) | DSR · Nguyen suite average | recovery rate: 83.6% vs 46.7% | Hurt |
| Train with risk-seeking policy gradients instead of standard policy gradients (ablation) | DSR on Nguyen-8 · Nguyen-8 | best-expression reward | Helped |
| Use standard policy gradients rather than risk-seeking gradients (ablation) | DSR on Nguyen-8 · Nguyen-8 | mean reward over the full batch | Helped |
| Add post hoc constraints to GP (ablation) | GP · Nguyen benchmarks | recovery rate: 60.1% to 32.3% | Hurt |
| Increase training dataset size by 10-fold (ablation) | DSR and baselines · Noisy Nguyen benchmarks | recovery rate | Mixed |
| Add Gaussian noise to the training targets (ablation) | DSR and baselines · Noisy Nguyen benchmarks | recovery rate | Mixed |
| Add Gaussian noise directly to the reward | DSR rather than PQT · Nguyen-4 | recovery rate | Helped |
| Replace risk-seeking gradients with priority queue training | DSR framework · Nguyen suite | recovery rate: 83.6% vs 75.2% | Hurt |
Closest papers by meaning
- A Unified Framework for Deep Symbolic Regression (NeurIPS 2022 · similarity 0.92)
- A Neural-Guided Dynamic Symbolic Network for Exploring Mathematical Expressions from Data (ICML 2024 · similarity 0.92)
- Deep Generative Symbolic Regression (ICLR 2023 · similarity 0.92)
- SMILE: Bridging Continuous Optimization and Discrete Symbolic Recovery (NeurIPS 2026 · similarity 0.89)
- Symbolic regression via MDLformer-guided search: from minimizing prediction error to minimizing description length (ICLR 2025 · similarity 0.89)
- SNIP: Bridging Mathematical Symbolic and Numeric Realms with Unified Pre-training (ICLR 2024 · similarity 0.88)
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.