Prof. Lei CHEN and His Students Received the Best Regular Research Paper at 48th International Conference on Very Large Databases

Prof. Lei CHEN, Chair Professor of Department of Computer Science & Engineering with his students and researchers, received the Best Regular Research paper for his paper "SANCUS: Staleness-Aware Communication-Avoiding Full-Graph Decentralized Training in Large-Scale Graph Neural Networks" at 48th International Conference on Very Large Databases in Sydney, Australia.

The award-winning paper co-authored with Jingshu PEN, Zhao CHEN, Yingxia SHAO, Yanyan SHEN and Jiannong CAO, propose to use SANCUS to abstracts decentralized Graph Neural Networks (GNN) processing as sequential matrix multiplication and uses historical embeddings via cache. Theoretically, the team shows bounded approximation errors of embeddings and gradients with convergence guarantee. Empirically, compared to SOTA works, SANCUS can avoid up to 74% communication with at least 1.86 times faster throughput on average without accuracy loss.

VLDB is a premier annual international forum for data management, scalable data science and database researchers, vendors, practitioners, application developers, and users. The VLDB 2022 conference program features with 252 research paper talks (selected from 949 submissions), eleven keynote & invited talks, two panels, nine tutorials, over 40 demonstrations, and 10 workshops, including the VLDB PhD Workshop. It covers issues in data management, database architectures, graph data management, data privacy and security, data mining, machine learning, AI and database systems research - all essential technological cornerstones of the emerging applications of the 21st century.

Congratulations again to Prof. CHEN, his students and researchers.

For more details, please refer to the VLDB2022 website.

Prof. Lei CHEN presented at 48th International Conference on Very Large Databases in Sydney, Australia

Highlight on the award-winning paper

Highlight on the award-winning paper

The "Best Regular Research paper" for "SANCUS: Staleness-Aware Communication-Avoiding Full-Graph Decentralized Training in Large-Scale Graph Neural Networks"