Evaluating Federated Learning Scenarios: Impact of Basic Parameters and Realistic Communication in Testbeds

• Bruno H. Meyer UFPR
• Aurora Pozo UFPR
• Michele Nogueira UFMG
• Wagner M. Nunan Zola UFPR

DOI: https://doi.org/10.5753/eradrs.2025.6837

Resumo

This paper evaluates Federated Learning in different scenarios to identify differences between simulations and hardware. The scenarios include the MENTORED Testbed with x86 nodes, an experimental cluster with aarch64 architectures, and a server simulating FL using Docker.

Referências

Božič, J., Faustino, A. R., Radovič, B., Canini, M., and Pejović, V. (2024). Where is the testbed for my federated learning research? In 2024 IEEE/ACM Symposium on Edge Computing (SEC), pages 249–264. IEEE.

Liu, B., Lv, N., Guo, Y., and Li, Y. (2024). Recent advances on federated learning: A systematic survey. Neurocomputing, page 128019.

Meyer, B. H., Pozo, et al. (2024). Aprendizado federado com autokeras e knowledge distillation. In Anais da XXIV Escola Regional de Alto Desempenho da Região Sul.