AttriReBoost: A Gradient-Free Propagation Optimization Method for Cold Start Mitigation in Attribute Missing Graphs

Mengran Li, İD, Chaojun Ding, İD, Junzhou Chen, İD, Wenbin Xing, İD, Cong Ye, İD, Ronghui Zhang, İD, İD, Jia Hu, İD, Tony Z. Qiu, İD, Huijun Gao, İD · 2025

Read AttriReBoost: A Gradient-Free Propagation Optimization Method for Cold Start Mitigation in Attribute Missing Graphs by Mengran Li, İD, Chaojun Ding, İD, Junzhou Chen, İD, Wenbin Xing, İD, Cong Ye, İD, Ronghui Zhang, İD, İD, Jia Hu, İD, Tony Z. Qiu, İD, Huijun Gao, İD (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-00743.txt.

Description

Mengran Li and a team of co-authors repair incomplete graphs in twenty twenty-five. Many real graphs lack attributes for some nodes. Graph neural networks trained on such data produce biased results. Feature propagation is the standard repair, filling gaps from neighbors. It suffers a cold start problem, worst for low-degree nodes and reset attributes.

Their method, AttriReBoost, rethinks how propagation begins. New boundary conditions and added virtual edges improve node connectivity. Propagation then converges more stably and efficiently. The reconstruction needs no gradient computation, so overhead stays low. Readers working with messy real-world graphs get a way to restore missing features without training a heavy imputation model first.

Catalog record

CREATORMengran Li, İD, Chaojun Ding, İD, Junzhou Chen, İD, Wenbin Xing, İD, Cong Ye, İD, Ronghui Zhang, İD, İD, Jia Hu, İD, Tony Z. Qiu, İD, Huijun Gao, İD
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE65 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-00743

Browse the full library · About the Tower of Babel