LEAF

LEAF is a benchmark dataset used to evaluate federated learning algorithms, containing various tasks designed to simulate the challenges of training machine learning models across distributed networks of mobile devices.

Papers using LEAF (8)

Auto-FL-Research: Agentic Search for Federated Learning Algorithms2026Communication-Efficient Federated Learning With Data and Client Heterogeneity2022Federated Meta-Learning with Fast Convergence and Efficient Communication2018 · 279 citesReconciling Security and Communication Efficiency in Federated Learning2022 · 4 citesFAT: Federated Adversarial Training2020 · 3 citesCurse or Redemption? How Data Heterogeneity Affects the Robustness of Federated Learning2021 · 3 citesTiFL: A Tier-based Federated Learning System2020 · 2 citesFedMPQ: Secure and Communication-Efficient Federated Learning with Multi-codebook Product Quantization2024