Federated Meta-Learning with Fast Convergence and Efficient Communication

2018/02/22 by Fei Chen, Mi Luo, Chen, Fei +7 · 10 citations
Computer Science · #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Internet Traffic Analysis and Secure E-voting #Machine Learning (cs.LG) #Privacy-Preserving Technologies in Data

paper · pdf · doi:10.48550/arxiv.1802.07876

openalex publication_date 2018/02/22 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Statistical and systematic challenges in collaboratively training machine learning models across distributed networks of mobile devices have been the bottlenecks in the real-world application of federated learning. In this work, we show that meta-learning is a natural choice to handle these issues, and propose a federated meta-learning framework FedMeta, where a parameterized algorithm (or meta-learner) is shared, instead of a global model in previous approaches. We conduct an extensive empirical evaluation on LEAF datasets and a real-world production dataset, and demonstrate that FedMeta achieves a reduction in required communication cost by 2.82-4.33 times with faster convergence, and an increase in accuracy by 3.23%-14.84% as compared to Federated Averaging (FedAvg) which is a leading optimization algorithm in federated learning. Moreover, FedMeta preserves user privacy since only the parameterized algorithm is transmitted between mobile devices and central servers, and no raw data is collected onto the servers.

Citations

• Learning to reinforcement learn
• RL²: Fast Reinforcement Learning via Slow Reinforcement Learning
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• On First-Order Meta-Learning Algorithms
• LEAF: A Benchmark for Federated Settings
• A Closer Look at Few-shot Classification
• Towards Federated Learning at Scale: System Design
• DeepFM: A Factorization-Machine based Neural Network for CTR Prediction

Cited by

• Federated Learning via Meta-Variational Dropout
• Parameter-Efficient and Personalized Federated Training of Generative Models at the Edge
• Ditto: Fair and Robust Federated Learning Through Personalization
• Federated Learning with Heterogeneous and Private Label Sets
• GraphFL: A Federated Learning Framework for Semi-Supervised Node Classification on Graphs
• A Field Guide to Federated Optimization
• Edge Intelligence: Architectures, Challenges, and Applications
• FedMeNF: Privacy-Preserving Federated Meta-Learning for Neural Fields
• Two Sides of Meta-Learning Evaluation: In vs. Out of Distribution
• Personalized Adaptive Meta Learning for Cold-start User Preference Prediction

Related

• Federated Learning and Meta Learning: Approaches, Applications, and Directions
• Improving Federated Learning Personalization via Model Agnostic Meta Learning
• Personalized Federated Learning with Contextual Modulation and Meta-Learning
• Federated Learning under Heterogeneous and Correlated Client Availability
• Elastically-Constrained Meta-Learner for Federated Learning