Winners of VLDB 2022 unveiled, Prof. Lei Chen and his team received the Best Regular Research Paper
Recently, Prof. Lei Chen, Acting Dean of Data Science and Analytics Thrust, Information Hub, HKUST(GZ) and Chair Professor of the Department of Computer Science and Engineering of HKUST, and his team 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 the International Conference on Very Large Data Bases (VLDB) 2022 held in Sydney, Australia.
VLDB, together with SIGMOD and ICD, represents three premier international conferences in the area of the database. Since launched in 1975, VLDB has been held for 48 sessions. The paper submission for VLDB 2022 includes four categories, namely Regular Research Papers, Scalable Data Science Papers, Experiment, Analysis & Benchmark Papers, and Vision Papers, with 949 papers submitted and 252 papers included. The three best research papers were selected from the categories of Regular Research, Experiment, Analysis & Benchmark, and Scalable Data Science respectively. The award that Prof. Lei Chen and his team won is the Best Regular Research Paper, which is the most influential.
During the conference, Prof. Lei Chen delivered an invited talk on behalf of the project team. He introduced the systematic research efforts of his team made in generalization models of graph neural networks (GNN). The winning of the award is the strong evidence of HKUST(GZ)’s influence in the international academic community in the area of data science.
To commend his outstanding contributions when serving as Editor-in-Chief of the VLDB In recent years, GNN has gained wide attention as an emerging graph data learning technique. It is learned that the award-winning paper was finished by the Beijing University of Posts and Telecommunications, Shanghai Jiao Tong University and Hong Kong Polytechnic University. The paper proposes a unique distributed training framework, SANCUS, and uses the decentralized mechanism to accelerate the distributed training of GNN to reducing communication. The paper has not only theoretically proved that the convergence speed approximates the full-graph training, but also verified the training efficiency and accuracy of SANCUS by the large number of experiments in real scenarios.
Prof. Lei Chen is an internationally renowned scholar in the field of the database, an IEEE Fellow and ACM Distinguished Scientist. He has been widely recognized for his academic achievements across the academia. He is serving as the Editor-in-Chief of the Transactions on Knowledge and Data Engineering (TKDE), a journal of IEEE focusing on data mining in computing, and the Journal Associate Editor of the VLDB Journal. He is also the co-chair of IEEE International Conference on Data Engineering (ICDE) 2023 and the secretary of the Executive Committee of the VLDB Foundation.
To commend his outstanding contributions when serving as Editor-in-Chief of the VLDB Journal from 2015 to 2022, the conference presented Prof. Lei Chen with the Service Award for his contributions to the development of database scholar community.
It is worth mentioning that Prof. Lei Chen and his team secured the success of the bid for hosting VLDB 2024 on behalf of HKUST(GZ). In 2024, the 50th VLDB will be held in HKUST(GZ) which sits in Nansha District, Guangzhou. It will be the second time VLDB held in a city in the Chinese mainland after Hangzhou, which hosted the VLDB 2014.