Research map / Probabilistic and scientific modeling / Molecules and proteins
Transformer-based Planning for 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 MCTS-guided decoding with fitting and complexity feedback instead of transformer sampling | E2E · Feynman | R^2>0.99: 0.848 -> 0.949 | Helped |
| Use MCTS-guided decoding with fitting and complexity feedback instead of transformer sampling | E2E · Strogatz | R^2>0.99: 0.357 -> 0.785 | Helped |
| Use MCTS-guided decoding with fitting and complexity feedback instead of transformer sampling | E2E · Black-box | R^2: 0.864 -> 0.945 | Helped |
| Use MCTS-guided decoding with fitting and complexity feedback instead of transformer sampling | E2E · In-domain Synthetic | Acc_0.01: 0.332 -> 0.326 | Hurt |
| Increase complexity reward weight from λ=0 to λ=0.1 (ablation) | TPSR · Feynman | Complexity: 84.42 -> 57.22 | Helped |
| Increase complexity reward weight from λ=0 to λ=0.1 (ablation) | TPSR · Feynman | R^2>0.99: 0.952 -> 0.949 | Hurt |
| Increase complexity reward weight from λ=0.1 to λ=1 (ablation) | TPSR · Feynman | Complexity: 57.22 -> 47.24 | Helped |
| Increase complexity reward weight from λ=0.1 to λ=1 (ablation) | TPSR · Feynman | R^2>0.99: 0.949 -> 0.916 | Hurt |
| Add sequence caching (ablation) | TPSR (λ=0.1) · Feynman | inference time | Helped |
| Add top-k caching (ablation) | TPSR (λ=0.1) · Feynman | inference time | Helped |
| Add both sequence and top-k caching (ablation) | TPSR (λ=0.1) · Feynman | inference time: around 28% | Helped |
| Increase the number of MCTS rollouts (ablation) | TPSR (λ=0.1) · Feynman | fitting performance | Helped |
Closest papers by meaning
- Improving Monte Carlo Tree Search for Symbolic Regression (NeurIPS 2025 · similarity 0.88)
- ODEFormer: Symbolic Regression of Dynamical Systems with Transformers (ICLR 2024 · similarity 0.87)
- SMILE: Bridging Continuous Optimization and Discrete Symbolic Recovery (NeurIPS 2026 · similarity 0.87)
- Symbolic Regression with a Learned Concept Library (NeurIPS 2024 · similarity 0.87)
- Symbolic regression via MDLformer-guided search: from minimizing prediction error to minimizing description length (ICLR 2025 · similarity 0.87)
- ParFam -- (Neural Guided) Symbolic Regression via Continuous Global Optimization (ICLR 2025 · similarity 0.87)
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.