摘要
由一位新闻记者兼机器人与机器学习每日新闻的工作人员新闻编辑，关于进化计算的最新研究结果已经发表。根据Ne wsRx编辑在法国图卢兹的新闻报道，研究表明：“多目标进化算法(MOEAs)的性能因问题而异，因此很难开发新的算法或将现有算法应用于新的问题。为了简化新的多目标算法的开发和应用，人们对它们从组件中自动设计的兴趣越来越大。”新闻编辑们从图卢兹大学的研究中引用了一句话：“这些自动设计的元启发式算法可以比人类开发的ed算法更好。”本文提出了一种新的方法来研究自动签名算法的最终配置对算法性能的影响,并将该方法应用于由迭代竞赛(I Race)配置包设计的基于分解(MOEA/D)的优化多目标进化算法,解决了3组约束问题:(1)分析现实世界问题，(2)解析人工问题和(3)模拟真实世界。然后我们比较了算法组件在搜索轨迹网络(STNs)、种群多样性和任意超体积值方面的影响。MOEAs研究D在搜索的一半之前收敛到分析性人工问题和分析性现实世界问题中普遍良好的HV值。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News-Current study results on evolutionary computation have been published. According to news reporting out of Toulouse, France, by Ne wsRx editors, research stated, "The performance of multiobjective evolutionary a lgorithms (MOEAs) varies across problems, making it hard to develop new algorith ms or apply existing ones to new problems. To simplify the development and appli cation of new multiobjective algorithms, there has been an increasing interest i n their automatic design from their components." The news editors obtained a quote from the research from University of Toulouse: "These automatically designed metaheuristics can outperform their human-develop ed counterparts. However, it is still unknown what are the most influential comp onents that lead to performance improvements. This study specifies a new methodo logy to investigate the effects of the final configuration of an automatically d esigned algorithm. We apply this methodology to a tuned Multiobjective Evolution ary Algorithm based on Decomposition (MOEA/D) designed by the iterated racing (i race) configuration package on constrained problems of 3 groups: (1) analytical real-world problems, (2) analytical artificial problems and (3) simulated real-w orld. We then compare the impact of the algorithm components in terms of their S earch Trajectory Networks (STNs), the diversity of the population, and the anyti me hypervolume values. Looking at the objective space behavior, the MOEAs studie d converged before half of the search to generally good HV values in the analyti cal artificial problems and the analytical real-world problems."
NETL
NSTL