Federated Learning with Non-IID Data

Explore this paper's citation graph

Summary

This work presents a strategy to improve training on non-IID data by creating a small subset of data which is globally shared between all the edge devices, and shows that accuracy can be increased by 30% for the CIFAR-10 dataset with only 5% globally shared data.

Type
preprint
Published
2018-06-02
Cited by
3,436
References
31
Access
Open access

Keywords

Computer science, Train, Focus (optics), Enhanced Data Rates for GSM Evolution, Federated learning

References

Cited by

Related papers