Abstract:Different from traditional supervised learning framework in which each obj ect is assigned to only one concept of label,the condition which one obj ect may be associated with multiple labels simultaneously in multi-label learning is able to analyze the problems in the real world more effectively.In recent years,multi-label learning has been attracting a great deal of attention in machine learning.In multi-label learning,each obj ect may be associated with multiple labels simultaneously,and these labels are related to each other.So how to effectively discover and exploit correlations among labels is the core research issue of multi-label learning.A series of multi-label learning al-gorithms by exploiting label correlations have been proposed and applied successfully in many application areas. However,there are a lot of redundant features and irrelevant features existing in high dimensional data which reduce the performance of classifiers,and few multi-label feature selection algorithms consider the label correlations. Meanwhile,most of the existed algorithms for multi-label feature selection exploit the global label correlations,which assuming that the label correlations are shared by all the instances.However,in real-world tasks,different instances may share different label correlations.With this respect,we focuse on how to exploit the label correlations locally to improve multi-label feature selection and help multi-label classification.In this paper,a novel multi-label feature selection algorithm by exploiting the label correlation locally (Loc-MLFS )is introduced.The algorithm takes advantages of local label correlations(the correlations are not shared by all instances)in multi-label feature selection algorithm.To achieve the use of local label correlations,Loc-MLFS divides the samples into groups by category clustering and use multi-label feature selection to each group.At the same time,the algorithm can be extended to a unified framework.Experimental results on the datasets demonstrate that Loc-MLFS achieves superior performance.

Learning Multi-Label Label-Specific Features Via Global and Local Label Correlations

Exploring Common and Label-Specific Features for Multi-Label Learning with Local Label Correlations

Multi-label Learning Based on Label-Specific Features and Local Pairwise Label Correlation.

Multi-Label Learning with Global and Local Label Correlation

Learning common and label-specific features for multi-Label classification with correlation information

Joint Label-Specific Features and Correlation Information for Multi-Label Learning.

Multi-Label Learning with Label Specific Features Using Correlation Information.

Enhancing Label Correlations in Multi-Label Classification Through Global-Local Label Specific Feature Learning to Fill Missing Labels

Multi-label Learning with Label-Specific Features by Resolving Label Correlations.

Global-Local Label Correlation for Partial Multi-Label Learning

Improving Multi-Label Learning by Modeling Local Label and Feature Correlations

Multi-Label Learning by Exploiting Label Correlations Locally

Learning Label-Specific Features for Multi-Label Classification with Missing Labels

Robust Multi-Label Classification with Enhanced Global and Local Label Correlation

Learning Label-Specific Multiple Local Metrics for Multi-Label Classification

Integrating Global and Local Feature Selection for Multi-Label Learning

Improving Multi-Label Classification with Missing Labels by Learning Label-Specific Features

Learning Common and Label-Specific Features for Multi-Label Classification With Missing Labels

Multi-label Local-to-Global Feature Selection

A multi-label feature selection algorithm by exploiting label correlations locally

K-Labelsets For Multimedia Classification With Global And Local Label Correlation