Quaternion adaptive approximation normalization graph guided implicit low rank for robust matrix completion

Yu Guo

doi:10.1016/j.patcog.2026.113210

Quaternion adaptive approximation normalization graph guided implicit low rank for robust matrix completion

2026-01-30

0

PRE

AI

DOI：10.1016/j.patcog.2026.113210

Original

Original request for help

Save

Abstract

En 中文

• We introduce a novel adaptive approximation normalization Laplacian. • It avoids matrix inverse operations and preserves the Laplacian’s symmetry by introducing only a single adaptive scalar. • We propose a new model, the quaternion adaptive approximate normalization graph (QAANG), which combines graph regularity and low-rankness. • QAANG model surpasses state-of-the-art quaternion methods in both performance and robustness.

Keywords:

adaptive approximation

Laplacian normalization

quaternion graph

low-rankness

matrix completion

Quaternion adaptive approximation normalization graph guided implicit low rank for robust matrix completion

Abstract

Journal

Organization