Problem Decomposition Strategies and Credit Distribution Mechanisms in Modular Genetic Programming for Supervised Learning
综述了70篇关于遗传规划中问题分解的文献，提出一个统一分类体系，从架构、分解策略和信用分配三个维度梳理现有方法，帮助研究者把握全局并识别未来研究方向。
In this review article, we provide a comprehensive guide to the endeavor of problem decomposition within the field of Genetic Programming (GP), specifically tree-based GP for supervised learning tasks. We analyzed in detail 70 manuscripts that deal with motifs such as “problem decomposition”“, modular GP”“, subroutine evolution”“, hierarchical GP”“, cooperative coevolution”, among others. As a result of this study, we propose an unifying taxonomy that categorizes efforts on problem decomposition in GP along three major axes: the architecture of evolved composite solutions, problem decomposition strategy, and credit assignment approach. This classification system sheds light on how the diverse proposed methodologies for problem decomposition relate to each other and where most of the research efforts have focused to this day. Rather than discussing in detail any particular set of works, we see this overview as a map that may help researchers in obtaining a wider view of existing efforts for problem decomposition in GP, as well as provide a cohesive framework that allows the disclosure of future developments in clearly differentiated niches. We close the article with a brief analysis that compares the current state of problem decomposition methodologies in GP with that of another exemplar of problem decomposition in machine learning: deep learning.