基于遗传编程的符号回归在化学和材料研究中的应用与展望  

Application and Prospects of Symbolic Regression Based on Genetic Programming inChemistry and Materials

作　　者：田嘉欣 李浩源 TIAN Jiaxin;LI Haoyuan(School of Microelectronics,Shanghai University,Shanghai 201800,China)

机构地区：[1]上海大学微电子学院,上海201800

出　　处：《材料导报》2024年第S01期268-274,共7页Materials Reports

基　　金：国家自然科学基金(22103053)。

摘　　要：化学和材料学科的理论进步可以促进化学反应设计、材料合成和工艺条件优化。目前,在许多前沿领域的复杂问题中,传统自底向上的推导理论面临诸多挑战,而基于遗传编程的符号回归方法在挖掘数据关系方面展示了独特的优势,为解决复杂问题提供了新的思路。本文主要综述了基于遗传编程的符号回归方法的原理、优势及局限性,讨论了其近年来的发展趋势,整理了简单易用的遗传编程工具,进而阐述了该方法在化学与材料领域的代表性应用,最后对遗传编程在材料与化学领域的未来发展方向进行了展望。Theoretical advancements in the disciplines of chemistry and materials science can promote the design of chemical reactions,the synthesis of materials,and the optimization of process conditions.Currently,traditional bottom-up deductive theories face many challenges in addressing complex problems in many cutting-edge fields.The symbolic regression method based on genetic programming has shown unique advantages in mining data relationships,offering new perspectives for solving complex problems.This article mainly reviews the principles,advantages,and limitations of the symbolic regression method based on genetic programming.It discusses its development trends in recent years,compiles easy-to-use genetic programming tools,and further elaborates on representative applications of the method in the fields of chemistry and materials.Finally,the article provides a perspective on the future development directions of genetic programming in the fields of materials and chemistry.

关 键 词：可解释机器学习模型 符号回归 遗传编程 性质预测 工艺条件优化 

分 类 号：TQ015.9[化学工程]