An improved binary grey wolf optimizer for constrained engineering design problems

计算机科学 特征选择 过度拟合 特征(语言学) 采样(信号处理) 人工智能 二进制数 算法 模式识别(心理学) 数学优化 数学 语言学 人工神经网络 算术 滤波器(信号处理) 哲学 计算机视觉
作者
Parijata Majumdar,Diptendu Bhattacharya,Sanjoy Mitra,Leonardo Ramos Rodrigues,Diego Oliva
出处
期刊:Expert Systems [Wiley]
标识
DOI:10.1111/exsy.13458
摘要

Abstract An Improved binary Non‐Linear Convergent Bi‐phase Mutated Grey Wolf Optimizer (IbGWO) is proposed for solving feature selection problems with two main goals reducing irrelevant features and maximizing accuracy. We used stratified ‐fold cross‐validation that performs stratified sampling on the data to avoid overfitting problems. The fitness function used in the proposed algorithm allows choosing the solution with the minimum number of features if more than one feature has the same highest accuracy. When stratified cross‐validation is performed, the split datasets contain the same share of the feature of interest as the actual dataset. During stratified sampling, the cross‐validation result minimizes the generalization error to a considerable extent, with a smaller variance. Feature selection could be seen as an optimization problem that efficiently removes irrelevant data from high‐dimensional data to reduce computation time and improve learning accuracy. This paper proposes an improved Non‐Linear Convergent Bi‐Phase Mutated Binary Grey Wolf Optimizer (IbGWO) algorithm for feature selection. The bi‐phase mutation enhances the rate of exploitation of GWO, where the first mutation phase minimizes the number of features and the second phase adds more informative features for accurate feature selection. A non‐linear tangent trigonometric function is used for convergence to generalize better while handling heterogeneous data. To accelerate the global convergence speed, an inertia weight is added to control the position updating of the grey wolves. Feature‐weighted K‐Nearest Neighbor is used to enhance classification accuracy, where only relevant features are used for feature selection. Experimental results confirm that IbGWO outperforms other algorithms in terms of average accuracy of 0.8716, average number of chosen features of 6.13, average fitness of 0.1717, and average standard deviation of 0.0072 tested on different datasets and in terms of statistical analysis. IbGWO is also benchmarked using unimodal, multimodal, and IEEE CEC 2019 functions, where it outperforms other algorithms in most cases. Three classical engineering design problems are also solved using IbGWO, which significantly outperforms other algorithms. Moreover, the overtaking percentage of the proposed algorithm is .

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
量子星尘发布了新的文献求助10
刚刚
梅川秋裤完成签到,获得积分10
1秒前
自由千风发布了新的文献求助10
1秒前
sai完成签到,获得积分10
1秒前
1秒前
三石完成签到 ,获得积分10
1秒前
帅气男孩发布了新的文献求助10
2秒前
meng发布了新的文献求助10
3秒前
安静的难破完成签到,获得积分10
3秒前
asdasd完成签到,获得积分10
4秒前
waypeter完成签到,获得积分10
5秒前
5秒前
大宝完成签到,获得积分10
6秒前
7秒前
苗条从雪完成签到,获得积分10
7秒前
李爱国应助Albertxkcj采纳,获得10
8秒前
lyz完成签到 ,获得积分10
8秒前
9秒前
CodeCraft应助科研通管家采纳,获得10
9秒前
所所应助科研通管家采纳,获得30
9秒前
852应助科研通管家采纳,获得10
9秒前
黄油板栗应助科研通管家采纳,获得10
9秒前
黄油板栗应助科研通管家采纳,获得10
9秒前
ding应助科研通管家采纳,获得10
9秒前
9秒前
米尔的猫应助科研通管家采纳,获得20
9秒前
9秒前
FashionBoy应助科研通管家采纳,获得10
9秒前
CodeCraft应助HelloKun采纳,获得10
10秒前
vvvv完成签到,获得积分10
10秒前
大模型应助微笑的桐采纳,获得10
11秒前
Emily完成签到,获得积分10
11秒前
单薄傲易发布了新的文献求助10
13秒前
忐忑的蓝血完成签到,获得积分10
16秒前
CC应助意去也采纳,获得10
17秒前
meng完成签到,获得积分10
17秒前
tiantian完成签到,获得积分10
18秒前
18秒前
FashionBoy应助lily采纳,获得30
18秒前
19秒前
高分求助中
A new approach to the extrapolation of accelerated life test data 1000
Cognitive Neuroscience: The Biology of the Mind 1000
Technical Brochure TB 814: LPIT applications in HV gas insulated switchgear 1000
ACSM’s Guidelines for Exercise Testing and Prescription, 12th edition 500
Picture Books with Same-sex Parented Families: Unintentional Censorship 500
Nucleophilic substitution in azasydnone-modified dinitroanisoles 500
不知道标题是什么 500
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 3969782
求助须知:如何正确求助?哪些是违规求助? 3514601
关于积分的说明 11174816
捐赠科研通 3249899
什么是DOI,文献DOI怎么找? 1795080
邀请新用户注册赠送积分活动 875599
科研通“疑难数据库(出版商)”最低求助积分说明 804886