A general convergence analysis method for evolutionary multi-objective optimization algorithm

维数之咒 插值(计算机图形学) 趋同(经济学) 数学优化 帕累托原理 多目标优化 维数(图论) 计算机科学 进化算法 算法 数学 人工智能 经济 经济增长 运动(物理) 纯数学
作者
Tie Cai,Hui Wang
出处
期刊:Information Sciences [Elsevier BV]
卷期号:663: 120267-120267 被引量:4
标识
DOI:10.1016/j.ins.2024.120267
摘要

Convergence analysis of multi-objective optimization algorithm has been an area of vital interest to the research community. With this regard, a number of approaches have been proposed and studied. However, these studies and developed proposals cannot cope with more than 3-dimensional optimization problems. Generally speaking, interpolation planes are formed by 3-dimension data. So, when the dimensionality of the Pareto front is more than 3, the dimensionality of Pareto front will be reduced to 3 by involving principal component analysis. This may lead to some important data being missed. Due to missing data, the formed interpolation plane is usually inaccurate and uneven. This will give rise to difficulties to evaluate the distance between the Pareto front and the optimal Pareto front. Subsequently, it is not easy to evaluate exact convergence time and with this regard the existing solutions lack general. Having this in mind, this paper develops a general convergence analysis (GCAM) for evolutionary multi-objective optimization algorithm (EMOA). In this approach, two originality aspects come to existence: one associates with the interpolation plane convergence analysis while the second concerns the improved drift analysis of evolutionary algorithm. Firstly, for more than 3-dimensional space, the dimensionality of the Pareto front set becomes reduced to 3 through a locally linear embedding. This overcomes the irregular interpolation plane problem and produce a high-quality interpolation. Secondly, this study originally analyzes the convergence of EMOA by engaging an improved drift analysis. Finally, we determine the first stopping time of EMOA by analyzing the convergence metric. The experimental results demonstrate that the proposed method exhibits better performance in comparison with CAD, CAL, and CAC. Specifically, the error proportion of SMS-EMOA, AR-MOEA, SPEA2+SDE, GFM-MOEA has been decreased by 12%, 15%, 21%, 19% and 17%, respectively.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
多经历经历完成签到,获得积分10
1秒前
张二十八完成签到,获得积分10
1秒前
1秒前
Deadman发布了新的文献求助10
1秒前
1秒前
错过的风景完成签到,获得积分10
2秒前
笨蛋小章发布了新的文献求助10
2秒前
追寻语芙完成签到,获得积分10
2秒前
科研通AI2S应助皇甫天佑采纳,获得10
2秒前
康康星发布了新的文献求助10
2秒前
SYMI驳回了桐桐应助
3秒前
量子星尘发布了新的文献求助10
4秒前
GLB完成签到,获得积分10
4秒前
4秒前
5秒前
奇尔韦尔发布了新的文献求助10
5秒前
6秒前
6秒前
叶明昭完成签到,获得积分10
7秒前
高工完成签到,获得积分10
7秒前
ddsdd发布了新的文献求助10
7秒前
8秒前
8秒前
8秒前
ycc完成签到,获得积分20
9秒前
掉头发的小白完成签到,获得积分10
9秒前
9秒前
默默的紫真完成签到,获得积分10
10秒前
LYD发布了新的文献求助10
10秒前
10秒前
10秒前
不懈奋进应助坦率续采纳,获得30
11秒前
666发布了新的文献求助10
11秒前
ww完成签到,获得积分10
11秒前
深情安青应助温暖南莲采纳,获得10
12秒前
wu完成签到,获得积分10
13秒前
13秒前
脑洞疼应助温水采纳,获得10
13秒前
顾矜应助dove采纳,获得10
14秒前
练习时长两年半应助XXXX采纳,获得10
14秒前
高分求助中
Picture Books with Same-sex Parented Families: Unintentional Censorship 700
ACSM’s Guidelines for Exercise Testing and Prescription, 12th edition 500
Nucleophilic substitution in azasydnone-modified dinitroanisoles 500
不知道标题是什么 500
Indomethacinのヒトにおける経皮吸収 400
Phylogenetic study of the order Polydesmida (Myriapoda: Diplopoda) 370
Effective Learning and Mental Wellbeing 300
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 遗传学 基因 物理化学 催化作用 冶金 细胞生物学 免疫学
热门帖子
关注 科研通微信公众号,转发送积分 3974712
求助须知:如何正确求助?哪些是违规求助? 3519159
关于积分的说明 11197254
捐赠科研通 3255257
什么是DOI,文献DOI怎么找? 1797724
邀请新用户注册赠送积分活动 877130
科研通“疑难数据库(出版商)”最低求助积分说明 806132