aiqbench analysis
MOSAIC-SR demonstrates how pretrained Transformers can effectively guide combinatorial search in symbolic regression, significantly improving the recovery of scientific equations from data. By leveraging learned priors to initialize promising structural regions, this method outperforms existing approaches in both solution rates and predictive accuracy across multiple benchmarks. The work highlights a critical advancement in machine learning for scientific discovery, bridging the gap between flexible neural inference and interpretable closed-form models.
MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery
arXiv:2609.20997v1 Announce Type: new Abstract: Symbolic regression aims to recover closed-form equations from observations, providing interpretable models for scientific discovery. Existing approaches struggle to combine flexible structural search with efficient inference. Search-based methods can refine expression structure but often rely on costly combinatorial optimization with random initialization. Pretrained neural models generate formulas almost instantly, but their predictions often contain symbolic errors. We introduce MOSAIC-SR, which uses a pretrained Transformer to propose multiple initial sketches. These sketches initialize searches in several promising regions, avoiding random starts in the vast expression space. Each search jointly recovers structure and constants through scale-aware constant optimization and local symbolic repair. We evaluate MOSAIC-SR on the SRSD-Feynman dataset with and without dummy variables and on six additional benchmarks. MOSAIC-SR obtains the highest symbolic solution rate on every dataset while ranking among the top two methods in predictive accuracy. This advantage persists in the presence of irrelevant dummy inputs. The results show that learned priors can focus search on promising equation structures, and that numerical optimization and symbolic repair are important for recovery.
Abstract:Symbolic regression aims to recover closed-form equations from observations, providing interpretable models for scientific discovery. Existing approaches struggle to combine flexible structural search with efficient inference. Search-based methods can refine expression structure but often rely on costly combinatorial optimization with random initialization. Pretrained neural models generate formulas almost instantly, but their predictions often contain symbolic errors. We introduce MOSAIC-SR, which uses a pretrained Transformer to propose multiple initial sketches. These sketches initialize searches in several promising regions, avoiding random starts in the vast expression space. Each search jointly recovers structure and constants through scale-aware constant optimization and local symbolic repair. We evaluate MOSAIC-SR on the SRSD-Feynman dataset with and without dummy variables and on six additional benchmarks. MOSAIC-SR obtains the highest symbolic solution rate on every dataset while ranking among the top two methods in predictive accuracy. This advantage persists in the presence of irrelevant dummy inputs. The results show that learned priors can focus search on promising equation structures, and that numerical optimization and symbolic repair are important for recovery.
Machine Learning (cs.LG)
arXiv:2609.20997 [cs.LG]
(or arXiv:2609.20997v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.20997 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 265 words
Related coverage
RISR: Residual-Informed Scientific Equation Discovery with Large Language Models
arXiv · Oct 10, 2026
ERBench: A Benchmark and Testsuite for Equation Discovery Algorithms
arXiv · Jun 9, 2026
AutoSR: Automatic Symbolic Regression by Searching Research States
arXiv · Aug 19, 2026
SRHarness: A Harness for Agentic Symbolic Regression
arXiv · Sep 30, 2026