.png){kind=icon}
返回
Feature selection using binary monarch butterfly optimization
.png){kind=icon}
PRE
AI
DOI：10.1007/s10489-022-03554-9.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Swarm intelligence algorithms have superior performance in searching for the optimal feature subset, where Monarch Butterfly Optimization (MBO) can solve the continuous optimization problem. However, there exist some defects for MBO such as the limited searchable positions, falling into local optimum easily and unsolved binary variables. To address these drawbacks, this paper develops two mechanisms to propose several revisions of binary MBO (BMBO) for metaheuristic feature selection. First, to make MBO suitable to solve the feature selection optimization problems, the S-shaped and V-shaped transfer functions are introduced to convert continuous space into binary, and then force the butterfly to move in the binary search space. Two updated positions of the monarch butterfly population are designed based on these above transfer functions respectively to construct two BMBO models, namely BMBO-S and BMBO-V, as the first mechanism of BMBO. Second, the new step length parameter is proposed to update the position of monarch butterfly individuals. To prevent MBO from falling into the local optimum, the local disturbance and group division strategies are added into MBO to construct new BMBO method. It follows that a mutation rate is employed to enhance the detection stage of BMBO, and the mutation operator-based BMBO (BMBO-M) is designed to avoid the premature convergence of MBO. Third, this fitness function is integrated with the KNN classifier and the weight of the feature subset length to rank the selected feature subset, and a metaheuristic feature selection algorithm with BMBO-M is developed. Experiments applied to nineteen low dimensional UCI datasets and seven high dimensional datasets demonstrate our designed algorithm has great classification efficiency when compared with the other related technologies.
Keyword:
Feature selection
Metaheuristic
Monarch butterfly optimization
Mutation operator
期刊
IF:
3.5
论文数:
7.6K
被引数:
1.7W
机构
引用论文
分享收藏Is there a role for the adrenals in the development of hypercholesterolemia in Zucker fatty rats?
PREAI
分享收藏Hybrid Multilabel Feature Selection Using BPSO and Neighborhood Rough Sets for Multilabel Neighborhood Decision Systems.png)
IEEE ACCESS
OAAI
IEEE ACCESSIF3.6
OAAI分享收藏Correlation-Guided Updating Strategy for Feature Selection in Classification With Surrogate-Assisted Particle Swarm Optimization代理辅助粒子群优化分类中特征选择的相关性引导更新策略
PREAI
分享收藏
Solving feature selection problems by combining mutation and crossover operations with the monarch butterfly optimization algorithm结合变异和交叉操作与monarch butterfly优化算法解决特征选择问题APPLIED INTELLIGENCE
APPLIED INTELLIGENCEIF3.5
PREAI
分享收藏
查看更多内容