南京大学学报（自然科学版）2025，Vol.61Issue(4)：645-659,15.DOI:10.13232/j.cnki.jnju.2025.04.010
面向缺失标签的基于数据依赖标签相关性多标签学习
Data-dependent dynamic label correlation learning for multi-label learning with missing label
摘要
Abstract
Multi-label learning assigns multiple labels to data instances,but real-world data often suffers from missing labels,increasing model complexity and prediction bias.Existing methods recover missing labels using predefined label correlations,yet neglect the compatibility between original label spaces and correlation matrices,introducing noise and spurious dependencies.To address this,we propose DDLC(Data-Dependent Dynamic Label Correlation Learning)method.By preserving label correlations through manifold regularization,DDLC employs a dynamic mapping function to recover the missing labels while suppressing the noise interference in the output space,adaptively adjusting the label associations across scenarios.Experiments on benchmark datasets demonstrate DDLC's superior performance and generalization capability.关键词
多标签学习/标签缺失/标签相关性/标签特定特征/近端梯度下降Key words
multi-label learning/missing label/label correlation/label-specific features/proximal gradient descent分类
信息技术与安全科学引用本文复制引用
蔡林晟,何卓新,毛煜,林耀进..面向缺失标签的基于数据依赖标签相关性多标签学习[J].南京大学学报（自然科学版）,2025,61(4):645-659,15.基金项目
国家自然科学基金(62076116) （62076116）
