Partial label zero-shot learning with semantic mining and instance–label alignment

doi:10.1016/j.engappai.2026.114854

Partial label zero-shot learning with semantic mining and instance–label alignment

2026-04-23

0

PRE

AI

陈晨 (C. L. Philip Chen)

DOI：10.1016/j.engappai.2026.114854

Original

Original request for help

Save

Abstract

En 中文

• A new framework for ZSL with ambiguous labels. • A semantic mining block detects noisy labels by instance-label matching. • A partial zero-shot loss reduces noise and aligns embeddings.

Keywords:

Zero-shot learning

Semantic mining

Label noise

Embedding alignment

Instance-label matching

Partial label zero-shot learning with semantic mining and instance–label alignment

Abstract

Journal

Organization