In a remarkable achievement, Ke Zhang, a senior computational scientist at ClusterTech, has made significant contributions to the field of research by publishing a groundbreaking paper. ClusterTech, a leading technology company specialising in high-performance computing and data analytics solutions, has been at the forefront of empowering organisations with advanced technologies. This recent accomplishment showcases the company's commitment to fostering innovation and driving progress in various scientific domains.
Authored by Ke Zhang and her fellows, the published paper, titled "Deep Efficient Private Neighbour Generation for Subgraph Federated Learning," delves into the emerging field of distributed subgraph federated learning. The paper was under the supervision of Professor Siu Ming Yiu at the University of Hong Kong, who specialises in cyber security, cryptography, and FinTech.
The paper proposes FedDEP to comprehensively tackle unique challenges in distributed subgraph federated learning (subgraph FL) brought by missing cross-subgraph neighbours. FedDEP consists of a series of novel technical designs: (1) Deep neighbour generation through leveraging the GNN embeddings of potential missing neighbours; (2) Efficient pseudo-FL for neighbour generation through em- bedding prototyping; and (3) Privacy protection through noise-less edge-local-differential-privacy. The paper analyses the correctness and efficiency of FedDEP, and provides theoretical guarantees on its privacy. Empirical results on four real-world datasets demonstrate the clear benefits of our proposed techniques.
FedDEP, proposed by Ke Zhang et al., offers a pioneering approach to address the challenges in distributed subgraph federated learning caused by missing cross-subgraph neighbours. Ke Zhang will represent ClusterTech at the ICDM-FedGraph* event in Shanghai from 30 Nov to 5 Dec, 2023. ClusterTech takes pride in supporting its talented staff members in driving innovation and providing cutting-edge solutions. The proposed technique has the potential to revolutionise the field of federated learning, enabling industries to achieve more accurate and comprehensive model training in various applications. ClusterTech's commitment to research and development ensures its continuous contribution to the advancement of the scientific community and industry at large.
About IEEE ICDM 2023:
The IEEE International Conference on Data Mining (ICDM) has established itself as the world’s premier research conference in data mining. It provides an international forum for presentation of original research results, as well as exchange and dissemination of innovative and practical development experiences. The conference covers all aspects of data mining, including algorithms, software, systems, and applications. ICDM draws researchers, application developers, and practitioners from a wide range of data mining related areas such as big data, deep learning, pattern recognition, statistical and machine learning, databases, data warehousing, data visualisation, knowledge-based systems, and high-performance computing. By promoting novel, high-quality research findings, and innovative solutions to challenging data mining problems, the conference seeks to advance the state-of-the-art in data mining.
https://www.cloud-conf.net/icdm2023/index.html
About FedGraph2023:
The FedGraph workshop aims to bring together researchers from different backgrounds with a common interest in how to extend current FL algorithms to operate with graph data models such as GNNs. FL is an extremely hot topic of large commercial interest and has been intensively explored for machine learning with visual and textual data. The exploration from graph mining researchers and industrial practitioners is timely catching up just recently. There are many unexplored challenges and opportunities, which urges the establishment of an organised and open community to collaboratively advance the science behind it. The prospective participants of this workshop will include researchers and practitioners from both graph mining and federated learning communities, whose interests include, but are not limited to: graph analysis and mining, heterogeneous network modelling, complex data mining, large-scale machine learning, distributed systems, optimisation, meta-learning, reinforcement learning, privacy, robustness, explainability, fairness, ethics, and trustworthiness.
https://sites.google.com/view/fedgraph2023
