南京大学学报（自然科学版）2024，Vol.60Issue(3)：511-522,12.DOI:10.13232/j.cnki.jnju.2024.03.014
基于模糊邻域熵的离群点检测方法
Fuzzy neighborhood entropy-based outlier detection
摘要
Abstract
Outlier detection(also known as anomaly detection)is an important research direction in the field of data mining,with the aim of identifying data points that are significantly different.In response to the problem of neglecting the fuzziness and neighborhood relationships of samples in outlier detection methods based on traditional rough set theory,this paper uses fuzzy neighborhood rough sets to compensate for the shortcomings of classical rough sets,and combines the uncertainty of entropy to propose a novel outlier detection method based on fuzzy neighborhood entropy.Firstly,a fuzzy neighborhood approximation space is constructed using fuzzy neighborhood radius and mixed fuzzy similarity.Then,a specific fuzzy neighborhood combination entropy and a relative fuzzy neighborhood combination entropy are defined to construct fuzzy neighborhood outliers.Furthermore,an outlier detection algorithm based on fuzzy neighborhood entropy(FNEOD)was designed by combining the outlier factor based on fuzzy neighborhood combination entropy.Finally,the FNEOD algorithm is compared with the main outlier detection algorithms.The experimental results show that this method has good effectiveness and adaptability.关键词
数据挖掘/离群点检测/模糊邻域组合熵/相对模糊邻域组合熵Key words
data mining/outlier detection/fuzzy neighborhood combination entropy/relative fuzzy neighborhood combination entropy分类
信息技术与安全科学引用本文复制引用
刘佳莉,陈锦坤..基于模糊邻域熵的离群点检测方法[J].南京大学学报（自然科学版）,2024,60(3):511-522,12.基金项目
国家自然科学基金(62076116,62076088),福建省自然科学基金(2020J01792,2021J02049) （62076116,62076088）
