Decouple-then-Align: Structural Consensus for Unsupervised Feature Selection

doi:10.1016/j.inffus.2026.104889

Decouple-then-Align: Structural Consensus for Unsupervised Feature Selection

2026-10-19

0

PRE

AI

DOI：10.1016/j.inffus.2026.104889

Original

Original request for help

Save

Abstract

En 中文

• DTA-UFS decouples graph preservation from the sparse-regression target. • A semi-orthogonal matrix aligns spectral and pseudo-label representations. • A conditional bound links consensus error to alignment and graph regularization.

Keywords:

Unsupervised feature selection

Procrustes alignment

Decouple-then-align

Embedding learning

Decouple-then-Align: Structural Consensus for Unsupervised Feature Selection

Abstract

Journal

Organization

Cited Papers