GESR: A Genetic Programming–Based Symbolic Regression Method with Gene Editing

Yanjie Li, Liping Zhang, Min Wu, Weijun Li, Lina Yu, Jingyi Liu, Yusong Deng, Mingzhu Wan, Xin Ning · 2026

Read GESR: A Genetic Programming–Based Symbolic Regression Method with Gene Editing by Yanjie Li, Liping Zhang, Min Wu, Weijun Li, Lina Yu, Jingyi Liu, Yusong Deng, Mingzhu Wan, Xin Ning (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-10685.txt.

Description

Yanjie Li and co-authors propose GESR, a genetic coding approach to symbol-based rule-finding that uses gene editing to tame random change and mixing. Symbol-based rule-finding seeks math laws from lab data, and genetic coding, which evolves formula shapes across rounds, remains a classical and widely used method. Its changes and mixings, however, are entirely random, producing good and bad shifts alike. GESR trains two BERT models as hands of God: one guides change of formula symbols through masked language models, and the other guides mixing by predicting the mixing point.

Tests show that GESR greatly improves run speed over old genetic coding methods and achieves strong overall performance across multiple symbol-based rule-finding tasks. By replacing blind shifts with guided editing, the method aims to keep evolution's searching power while cutting wasted shifts, offering a faster route to clear math formulas from data.

Catalog record

CREATORYanjie Li, Liping Zhang, Min Wu, Weijun Li, Lina Yu, Jingyi Liu, Yusong Deng, Mingzhu Wan, Xin Ning
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50650
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.10685

Browse the full library · About the Tower of Babel