Bayesian Nonparametric Federated Learning of Neural Networks

Abstract

In federated learning problems, data is scattered across different servers and exchanging or pooling it is often impractical or prohibited. We develop a Bayesian nonparametric framework for federated learning with neural networks. Each data server is assumed to provide local neural network weights, which are modeled through our framework. We then develop an inference approach that allows us to synthesize a more expressive global network without additional supervision, data pooling and with as few as a single communication round. We then demonstrate the efficacy of our approach on federated learning problems simulated from two popular image classification datasets.

ICML 2019

Cited by in corpus (16)

• A Field Guide to Federated Optimization
• A Federated Learning Aggregation Algorithm for Pervasive Computing: Evaluation and Comparison
• Federated Learning Based on Dynamic Regularization
• IBM Federated Learning: an Enterprise Framework White Paper V0.1
• Federated Learning with Matched Averaging
• FedMix: Approximation of Mixup under Mean Augmented Federated Learning
• No Fear of Heterogeneity: Classifier Calibration for Federated Learning with Non-IID Data
• Practical Federated Gradient Boosting Decision Trees
• SemiFed: Semi-supervised Federated Learning with Consistency and Pseudo-Labeling
• A Systematic Literature Review on Federated Learning: From A Model Quality Perspective
• Model-Contrastive Federated Learning
• Adversarial training in communication constrained federated learning
• Personalized Federated Learning with Gaussian Processes
• Hybrid Federated Learning: Algorithms and Implementation
• FedHealth 2: Weighted Federated Transfer Learning via Batch Normalization for Personalized Healthcare
• Federated CycleGAN for Privacy-Preserving Image-to-Image Translation