Federated Learning (FL) systems evolve in heterogeneous and ever-evolving environments that challenge their performance. Under real deployments, the learning tasks of clients can also evolve with time, which calls for the integration of methodologies such as Continual Learning. To enable research reproducibility, we propose a set of experimental best practices that precisely capture and emulate complex learning scenarios. Our framework, Freddie, is the first entirely configurable framework for Federated Continual Learning (FCL), and it can be seamlessly deployed on a large number of machines thanks to the use of Kubernetes and containerization. We demonstrate the effectiveness of Freddie on two use cases, (i) large-scale FL on CIFAR100 and (ii) heterogeneous task sequence on FCL, which highlight unaddressed performance challenges in FCL scenarios.
LibContinual: A Comprehensive Library towards Realistic Continual Learning
State Key Laboratory for Novel Software Technology, Nanjing University, China(南京大学新型软件技术国家重点实验室);School of Computing and Information Technology, University of Wollongong, Australia(沃林戈大学计算与信息科技学院);Department of Computer Science, University of Rochester, USA(罗切斯特大学计算机科学系)
Avalanche: an End-to-End Library for Continual Learning
University of Pisa(比萨大学);University of Bologna(博洛尼亚大学);Rochester Institute of Technology(罗切斯特理工学院);KU Leuven(鲁汶大学);Universitat Autònoma de Barcelona(巴塞罗那自治大学);Sapienza University of Rome(罗马第一大学);Baylor College of Medicine(贝勒医学院);Goethe University(歌德大学);ByteDance AI Lab(字节跳动人工智能实验室);University of California(加利福尼亚大学);New York University(纽约大学);Université Paris-Saclay(巴黎-萨克雷大学);University of Modena and Reggio-Emilia(摩德纳-雷焦艾米利亚大学);University of Hamburg(汉堡大学);Oxford Brookes University(牛津布鲁克斯大学);Orobix(Orobix公司);Numenta(Numenta公司);Scuola Normale Superiore(比萨高等师范学校)