The CUR Decomposition of Self-Attention Matrices in Vision Transformers
Lists
Only you see this
Abstract
Transformers have achieved great success in natural language processing and computer vision. The core and basic technique of transformers is the self-attention mechanism. The vanilla self-attention mechanism has quadratic complexity, which limits its
BibTeX
@article{wu2026cur,
author = {Chong Wu and Maolin Che and Hong Yan},
title = {{The CUR Decomposition of Self-Attention Matrices in Vision Transformers}},
journal = {IEEE Trans. Pattern Anal. Mach. Intell.},
year = {2026},
doi = {10.1109/tpami.2025.3646452},
}Loading related content...