Research map / Probabilistic and scientific modeling / Molecules and proteins
Deep Generative Symbolic Regression
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 DGSR instead of NGGP | symbolic regression · Feynman (d=5) | recovery rate (%): 63.44 $\pm$ 6.64 -> 69.69 $\pm$ 3.38 | Helped |
| Use DGSR instead of NGGP | symbolic regression · Synthetic (d=12) | recovery rate (%): 28.57 $\pm$ 0.00 -> 37.86 $\pm$ 5.62 | Helped |
| Use DGSR instead of NGGP | symbolic regression · Additional Feynman | recovery rate (%): 67.81 $\pm$ 3.00 -> 67.81 $\pm$ 4.60 | No clear effect |
| Use DGSR instead of NGGP | symbolic regression · Feynman (d=2) | recovery rate (%): 85.71 $\pm$ 0.00 -> 85.36 $\pm$ 0.69 | No clear effect |
| Add an encoder to the pretrained DGSR decoder (ablation) | DGSR · Feynman (d=5) | recovery rate (%): 66.66 -> 67.50 | Helped |
| Add pre-training to the decoder without an encoder (ablation) | DGSR · Feynman (d=5) | recovery rate (%): 60.41 -> 66.66 | Helped |
| Use NMSE rather than cross-entropy for pre-training (ablation) | DGSR · Feynman (d=5) | recovery rate (%): 62.50 -> 67.50 | Helped |
| Pre-train on d=5 rather than d=2 (ablation) | DGSR · Feynman (d=5) | recovery rate (%): 61.29 -> 67.50 | Helped |
| Use DGSR with the alternate policy-gradient optimizer, without the GP component (ablation) | DGSR · Feynman (d=2) | equation evaluations: 151,231 -> 29,356 | Helped |
Closest papers by meaning
- Contemporary Symbolic Regression Methods and their Relative Performance (NeurIPS 2021 · similarity 0.92)
- Deep symbolic regression: Recovering mathematical expressions from data via risk-seeking policy gradients (ICLR 2021 · similarity 0.92)
- Neural Symbolic Regression that scales (ICML 2021 · similarity 0.92)
- A Neural-Guided Dynamic Symbolic Network for Exploring Mathematical Expressions from Data (ICML 2024 · similarity 0.90)
- SNIP: Bridging Mathematical Symbolic and Numeric Realms with Unified Pre-training (ICLR 2024 · similarity 0.89)
- SMILE: Bridging Continuous Optimization and Discrete Symbolic Recovery (NeurIPS 2026 · similarity 0.89)
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.