A Performance Evaluation of Federated Learning Algorithms
Paper i proceeding, 2018

Federated learning is an approach to distributed machine learning where a global model is learned by aggregating models that have been trained locally on data-generating clients. Contrary to centralized optimization, clients can be very large in number and face challenges of data and network heterogeneity. Examples of clients include smartphones and connected vehicles, which highlights the practical relevance of federated learning. We benchmark three federated learning algorithms and compare their performance against a centralized approach where data resides on the server. The algorithms Federated Averaging (FedAvg), Federated Stochastic Variance Reduced Gradient, and CO-OP are evaluated on the MNIST dataset, using both i.i.d. and non-i.i.d. partitionings of the data. Our results show that FedAvg achieves the highest accuracy among the federated algorithms, regardless of how data was partitioned. Our comparison between FedAvg and centralized learning shows that they are practically equivalent when i.i.d. data is used. However, the centralized approach outperforms FedAvg with non-i.i.d. data.

Machine learning

Federated learning

Algorithm evaluation

Författare

Adrian Nilsson

Stiftelsen Fraunhofer-Chalmers Centrum för Industrimatematik

Simon Smith

Stiftelsen Fraunhofer-Chalmers Centrum för Industrimatematik

Gregor Ulm

Stiftelsen Fraunhofer-Chalmers Centrum för Industrimatematik

Emil Gustavsson

Stiftelsen Fraunhofer-Chalmers Centrum för Industrimatematik

Mats Jirstrand

Stiftelsen Fraunhofer-Chalmers Centrum för Industrimatematik

DIDL'18: PROCEEDINGS OF THE SECOND WORKSHOP ON DISTRIBUTED INFRASTRUCTURES FOR DEEP LEARNING

Vol. December 2018 1-8
978-1-4503-6119-4 (ISBN)

2nd Workshop on Distributed Infrastructures for Deep Learning (DIDL)
Rennes, France,

BADA - On-board Off-board Distributed Data Analytics

VINNOVA (2016-04260), 2016-12-01 -- 2019-12-31.

Ämneskategorier

Annan data- och informationsvetenskap

Bioinformatik (beräkningsbiologi)

Datorseende och robotik (autonoma system)

DOI

10.1145/3286490.3286559

Mer information

Senast uppdaterat

2020-01-16