融合差分教学优化的粗糙集属性约简算法  

Rough set attribute reduction algorithm based on differentialteaching-learning optimization

作　　者：周婉婷 郑颖春[1] 魏博涛 Zhou Wanting;Zheng Yingchun;Wei Botao(School of Science,Xi’an University of Science&Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学理学院,西安710054

出　　处：《计算机应用研究》2024年第11期3317-3322,共6页Application Research of Computers

基　　金：国家自然科学基金资助项目(12001420)。

摘　　要：针对传统粗糙集理论在属性约简中存在计算复杂度高、易陷入局部最优解等问题,结合差分教学优化算法的全局搜索能力和粗糙集在处理不精确和不确定数据方面的优势,提出融合差分教学优化的粗糙集属性约简算法(rough set attribute reduction algorithm based on differential teaching-learning optimization, AR-DTLBO)。首先,引入自适应教学因子和差分变异策略对教学优化算法进行改进,提高算法的搜索能力和优化性能;其次,通过改进后的教学优化算法“教”和“学”两个阶段对属性约简过程进行优化,降低了数据集的维度和复杂性;最后,在UCI数据库中的8个数据集上将所提算法和其他六种算法进行对比实验。实验结果表明,该算法在约简长度、约简时间、约简率和分类精度上均取得了良好的效果,实现了对数据集的有效约简和优化,能够有效减少冗余信息并提高决策规则的准确性,为决策分析和数据挖掘等领域提供了有效支撑。To address the challenges of high computational complexity and the tendency to get stuck in local optima during attribute reduction within traditional rough set theory,this paper proposed an innovative rough set attribute reduction algorithm based on differential teaching-learning optimization(AR-DTLBO).Leveraging the global search capabilities of the differential teaching-learning optimization algorithm along with the strengths of rough set theory in handling imprecise and uncertain data,the algorithm aimed to optimize the process.Firstly,it enhanced the teaching-learning optimization algorithm by introducing an adaptive teaching factor and a differential mutation strategy,thereby enhancing its search capabilities and optimization perfor-mance.Subsequently,it refined the attribute reduction process through the improved teaching-learning optimization algorithm’s“teaching”and“learning”phases,effectively reducing the dimensionality and complexity of datasets.Finally,it conducted comparative experiments between the proposed AR-DTLBO algorithm and six other algorithms,using eight datasets from the UCI database.The experimental results demonstrate that the proposed algorithm achieves favorable outcomes in terms of reduction length,reduction time,reduction rate,and classification accuracy.This successful reduction and optimization of datasets not only reduces redundant information but also enhances the precision of decision rules.These findings provide valuable support for decision analysis,data mining,and other related fields.

关 键 词：教学优化算法 教学阶段 学习阶段 差分变异策略 属性约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]