寻找覆盖各类进化算法选择方法的综述文献与博文

Hey there! Great question—tracking down surveys that cover beyond the "big name" selection methods in evolutionary algorithms (EAs) can feel like searching for a needle in a haystack, but I’ve got some reliable resources to share that should cover what you’re looking for:

Academic Surveys & Books #

• 《Evolutionary Algorithms: A Unified Approach》 (by Kenneth De Jong): This classic textbook dedicates an entire chapter to selection mechanisms, and it doesn’t stop at the mainstream methods. It dives deep into Boltzmann selection’s temperature-dependent fitness scaling, steady-state selection’s population update rules, and even touches on reward-based selection variants used in co-evolutionary systems.
• 《A Comprehensive Survey of Selection Methods in Evolutionary Computation》 (2018, Journal of Artificial Intelligence Research): This paper explicitly categorizes selection strategies into multiple groups, including "temperature-adjusted" (Boltzmann), "population-maintenance" (steady-state), and "feedback-driven" (reward-based) methods. It compares their strengths, use cases, and even includes empirical performance benchmarks across different EA frameworks.
• 《Handbook of Evolutionary Computation》 (edited by Thomas Bäck et al.): This authoritative handbook has dedicated sections on niche selection mechanisms. Look for entries on "adaptive selection strategies" and "co-evolutionary selection"—they cover reward-based selection in collaborative EA scenarios and break down steady-state selection’s differences from generational selection.

High-Quality Blog Posts & Course Materials #

• Deep Dive into EA Selection Strategies (Series) : Several independent computational science blogs have published multi-part series on EA selection. One standout series includes two detailed posts: one on Boltzmann selection’s practical implementation (including temperature decay schedules) and another that contrasts steady-state selection with truncation/rank-based methods, plus a section on reward-based selection for multi-objective optimization tasks.
• University Course Lectures: Public course materials from institutions like Stanford and the University of Cambridge include extended lectures on non-mainstream EA selection methods. For example, Cambridge’s Advanced Evolutionary Computation lecture notes have a module on adaptive selection that covers Boltzmann and reward-driven strategies, with code snippets (in Python) to demonstrate how they work.

Pro Tips for Further Searching #

• If you’re focused on specific use cases (e.g., combining EAs with reinforcement learning), try searching for keywords like reward-based selection evolutionary reinforcement learning—many recent research papers include mini-surveys in their introduction sections that summarize related work in this subfield.
• On academic databases, use the query: "selection methods" evolutionary algorithms survey Boltzmann steady-state reward-based to filter surveys that explicitly mention the methods you’re interested in.

内容的提问来源于stack exchange，提问作者Curious