Towards Taming the Resource and Data Heterogeneity in Federated Learning

By Zheng Chai, Hannan Fayyaz, Zeshan Fayyaz, Ali Anwar, Yi Zhou, Nathalie Baracaldo, Heiko Ludwig, Yue Cheng. 2019

94 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Li Li
2. Heiko Ludwig
3. Yuxi Fan
4. Nathalie Baracaldo
5. Kuo-Yi Lin
6. Yujing Chen
7. Yi Zhou
8. Ali Anwar

Citations in the corpus

• IBM Federated Learning: an Enterprise Framework White Paper V0.1 (2020, background): “This includes differences in distribution of data among participants, different dataset sizes for each party, different attributes and general issues of heterogeneity [11].”
• Asynchronous Online Federated Learning for Edge Devices with Non-IID Data (2019, background): “Since the server aggregates after all clients finish, the extended period of waiting time in a synchronized optimization protocol will lead to idling and wastage of computing resources [29], [30].”
• Asynchronous Online Federated Learning for Edge Devices (2019, background): “Since the server aggregates a‰er all clients €nish, the extended period of waiting time in a synchronized optimization protocol will lead to idling and wastage of computing resources [9, 10].”
• TiFL: A Tier-based Federated Learning System (2020, methodology): “To experimentally verify the above analysis and demonstrate the impact of resource heterogeneity and data quantity heterogeneity, we conduct a study with a setup similar to the paper [8].”
• A Survey on Federated Learning for Resource-Constrained IoT Devices (2021, unknown): “If the number of such straggler clients is high, then the overall model convergence would be at stake [99], [100].”
• A Survey on federated learning* (2020, background): “considered the impact of resource heterogeneity on the training time of FL [36].”
• Federated Learning in Smart City Sensing: Challenges and Opportunities (2020, unknown): “In addition, under the premise that the data characteristics of each client are aligned, the global model of Federated Learning can obtain the same modeling effect as the centralized storage of data [127].”