讲座题目：Towards Practical and Efficient Federated Learning

讲座时间：2023年6月28日 16:30-17:30

讲座人：Ahmed M. Abdelmoniem 助理教授

地点：T5-204

讲座内容：

Federated Learning (FL) enables distributed training by learners using local data, thereby enhancing privacy and reducing communication. However, it presents numerous challenges relating to the heterogeneity of the data distribution, device capabilities, and participant availability as deployments scale, which can impact both model convergence and bias. We show, via extensive empirical evaluation, that heterogeneity can negatively impact the performance and quality of FL-trained models. Then, we show that by adopting quantized-model training, the impact of heterogeneity can be mitigated. We also systematically try to address the question of resource efficiency in FL, showing the benefits of intelligent participant selection and incorporating updates from straggling participants. We demonstrate how these factors enable resource efficiency while also improving trained model quality. Then, we will briefly summarise current open problems in FL.

讲座人简介：

Ahmed M. Abdelmoniem is a Lecturer (Assistant Professor) at the School of Electronic Engineering and Computer Science, Queen Mary University of London, UK and Assiut University, Egypt and leads SAYED Systems Group (https://sayed-sys-lab.github.io/). He held the positions of Research Scientist at KAUST, Saudi Arabia, and Senior Researcher with Huawei's Future Networks Lab (FNTL), Hong Kong. He is an investigator on several national and international research projects totalling more than USD 650K in funding. He received his PhD in Computer Science and Engineering from the Hong Kong University of Science and Technology, Hong Kong in 2017. He was awarded the prestigious Hong Kong PhD. Fellowship from the Research Grant Council (RGC) of Hong Kong in 2013 to pursue his PhD at HKUST and a Competitive Research Grant at KAUST in 2021. He has published numerous papers in top venues and journals in the areas of distributed systems, computer networking, and machine learning. His current research interests are in the areas of optimizing systems supporting distributed machine learning, federated learning, and cloud/data-center networking with an emphasis on performance, practicality, and scalability.