CASIA OpenIR  > 学术期刊  > 自动化学报
基于相对离群因子的标签噪声过滤方法
侯森寓; 姜高霞; 王文剑
Source Publication自动化学报
ISSN0254-4156
2024
Volume50Issue:1Pages:154-168
Abstract分类任务中含有类别型标签噪声是传统数据挖掘中的常见问题,目前还缺少针对性方法来专门检测类别型标签噪声.离群点检测技术能用于噪声的识别与过滤,但由于离群点与类别型标签噪声并不具有一致性,使得离群点检测算法无法精确检测分类数据集中的标签噪声.针对这些问题,提出一种基于离群点检测技术、适用于过滤类别型标签噪声的方法 ——基于相对离群因子(Relative outlier factor, ROF)的集成过滤方法 (Label noise ensemble filtering method based on relative outlier factor, EROF).首先,通过相对离群因子对样本进行噪声概率估计;然后,再迭代联合多种离群点检测算法,实现集成过滤.实验结果表明,该方法在大多数含有标签噪声的数据集上,都能保持优秀的噪声识别能力,并显著提升各种分类模型的泛化能力.
Keyword分类 标签噪声 离群点检测 相对离群因子 噪声过滤
DOI10.16383/j.aas.c230117
Citation statistics
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/55761
Collection学术期刊_自动化学报
Recommended Citation
GB/T 7714
侯森寓,姜高霞,王文剑. 基于相对离群因子的标签噪声过滤方法[J]. 自动化学报,2024,50(1):154-168.
APA 侯森寓,姜高霞,&王文剑.(2024).基于相对离群因子的标签噪声过滤方法.自动化学报,50(1),154-168.
MLA 侯森寓,et al."基于相对离群因子的标签噪声过滤方法".自动化学报 50.1(2024):154-168.
Files in This Item: Download All
File Name/Size DocType Version Access License
AAS-CN-2023-0117.pdf(6186KB)期刊论文出版稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[侯森寓]'s Articles
[姜高霞]'s Articles
[王文剑]'s Articles
Baidu academic
Similar articles in Baidu academic
[侯森寓]'s Articles
[姜高霞]'s Articles
[王文剑]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[侯森寓]'s Articles
[姜高霞]'s Articles
[王文剑]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: AAS-CN-2023-0117.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.