Anchor-based graph embedding and soft label learning for multi-label classification with missing label

doi:10.1016/j.eswa.2025.129019

Anchor-based graph embedding and soft label learning for multi-label classification with missing label

2025-07-16

0

PRE

AI

DOI：10.1016/j.eswa.2025.129019

Original

Original request for help

Save

Abstract

En 中文

• We propose a novel multi-label missing label classification method, combining bipartite graph construction and semantic reconstruction. • This approach allows for explicit soft label predictions and implicit label space embedding under the sample consistency assumption. • We leverage high-rank and high-order label correlation information to learn representative label selection and semantic reconstruction. • Experiments on multi-label datasets from 12 benchmarks demonstrate that our approach achieves competitive results on the multi-label missing label task.

Keywords:

multi-label classification

missing labels

bipartite graph

semantic reconstruction

label correlation

Anchor-based graph embedding and soft label learning for multi-label classification with missing label

Abstract

Journal

Organization