Tag: #symbolic-regression5 results
ViSA-R2 recovers analytical SymPy expressions from 2D linear steady-state field visualizations using a self-verifying chain-of-thought pipeline mimicking physicist reasoning. It introduces ViSA-Bench, a synthetic benchmark with 30 verifiable scenarios for VLM evaluation. Built on 8B Qwen3-VL, it outperforms open-source baselines and frontier VLMs.
SymLang is a new framework integrating symmetry-constrained grammars, language-model-guided program synthesis, and Bayesian model selection to discover governing equations from noisy, partial observations. It achieves 83.7% exact structural recovery across 133 dynamical systems under 10% noise, outperforming baselines by 22.4 points. The open-source tool reduces extrapolation errors by 61% and minimizes conservation violations.
A novel LLM-based evolutionary search framework introduces programmatic context augmentation for symbolic regression. It enables code-based dataset interactions to extract signals beyond MSE metrics. Outperforms baselines on LLM-SRBench in efficiency and accuracy.
This discussion explores whether Symbolic Regression (SR) remains relevant given the rapid advancements in LLMs' code generation capabilities. It questions if LLMs can replace traditional SR techniques for model discovery.
PiT-PO uses reinforcement learning to evolve LLMs for symbolic regression, enforcing physical validity and parsimony. It treats LLMs as adaptive generators updated by search feedback. Achieves SOTA on benchmarks and discovers novel turbulence models.