A single-author preprint re-frames GNN over-smoothing as Anderson localization, defining a participation-degree metric and proposing degree-dependent edge reweighting as mitigation, without proof or experiments.
Spin glasses: Experimental facts, theoretical concepts, and open questions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Rethinking Over-Smoothing in Graph Neural Networks: A Perspective from Anderson Localization
A single-author preprint re-frames GNN over-smoothing as Anderson localization, defining a participation-degree metric and proposing degree-dependent edge reweighting as mitigation, without proof or experiments.