摘要
多标签分类是一项重要且具有挑战性的任务,对于场景分类、信息检索和网页挖掘等领域有重要意义.近年来,研究者倾向于挖掘并利用标签相关性以及实例相关性来提升多标签分类的性能,其中利用这类相关性来解决多标签分类中的标签缺失问题一直是机器学习领域的热门研究点,但现有的方法都忽略了标签和实例之间的相关性.针对标签完整和标签缺失的情况,基于矩阵分解的方式分解得到实例和标签的特征表示矩阵,在利用标签相关性和实例相关性建立正则约束的同时,首次建立实例和标签特征表示间的相关性并用以建立正则约束,进而提升算法性能.在三个真实数据集上对提出的算法进行性能验证,实验结果表明,该算法和对比算法相比,获得了最高的预测精度.
Multi-label classification is a challenging classification task,which is important in the fields of scene classification,information retrieval,and web mining.Recently,researchers tend to mine and use label correlation and instance correlation to improve the performance of multi-label classification.Using such correlation to solve the problem of missing labels in multilabel classification has been a research hot spot,but existing methods ignore the correlation between labels and instances.In this paper,aiming at the situation of complete labels and missing labels,the feature representation matrix of instance and labels is obtained based on matrix factorization.While using label correlation and instance correlation to establish constraints,the correlation between instance and label is firstly established.We validate the performance on three datasets and the proposed algorithm is superior to the existing methods.
作者
田小瑜
秦永彬
黄瑞章
陈艳平
Tian Xiaoyu;Qin Yongbin;Huang Ruizhang;Chen Yanping(State Key Laboratory of Public Big Data,Guizhou University,Guiyang,550025,China;The College of Computer Science and Technology,Guizhou University,Guiyang,550025,China)
出处
《南京大学学报（自然科学版）》
CAS
CSCD
北大核心
2023年第1期76-84,共9页
Journal of Nanjing University（Natural Science）
基金
国家自然科学基金(62066008)。
关键词
标签相关性
实例相关性
标签与实例相关性
矩阵分解
多标签分类
label correlation
instance correlation
correlation between instance and label
matrix factorization
multi-label classification
