Efficient Decentralized Deep Learning by Dynamic Model Averaging

Michael Kamp, Linara Adilova, Joachim Sicking, Fabian Hüger, Peter Schlicht, Tim Wirtz, Stefan Wrobel

ECML-PKDD 2018 pp. 393-409

doi:10.1007/978-3-030-10925-7_24 /ecmlpkdd/2018/kamp2018ecmlpkdd-efficient/

Abstract

We propose an efficient protocol for decentralized training of deep neural networks from distributed data sources. The proposed protocol allows to handle different phases of model training equally well and to quickly adapt to concept drifts. This leads to a reduction of communication by an order of magnitude compared to periodically communicating state-of-the-art approaches. Moreover, we derive a communication bound that scales well with the hardness of the serialized learning problem. The reduction in communication comes at almost no cost, as the predictive performance remains virtually unchanged. Indeed, the proposed protocol retains loss bounds of periodically averaging schemes. An extensive empirical evaluation validates major improvement of the trade-off between model performance and communication which could be beneficial for numerous decentralized learning applications, such as autonomous driving, or voice recognition and image classification on mobile phones.

PDF ECML-PKDD Semantic Scholar

Cite

Text

Kamp et al. "Efficient Decentralized Deep Learning by Dynamic Model Averaging." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2018. doi:10.1007/978-3-030-10925-7_24

Markdown

[Kamp et al. "Efficient Decentralized Deep Learning by Dynamic Model Averaging." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2018.](https://mlanthology.org/ecmlpkdd/2018/kamp2018ecmlpkdd-efficient/) doi:10.1007/978-3-030-10925-7_24

BibTeX

@inproceedings{kamp2018ecmlpkdd-efficient,
  title     = {{Efficient Decentralized Deep Learning by Dynamic Model Averaging}},
  author    = {Kamp, Michael and Adilova, Linara and Sicking, Joachim and Hüger, Fabian and Schlicht, Peter and Wirtz, Tim and Wrobel, Stefan},
  booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
  year      = {2018},
  pages     = {393-409},
  doi       = {10.1007/978-3-030-10925-7_24},
  url       = {https://mlanthology.org/ecmlpkdd/2018/kamp2018ecmlpkdd-efficient/}
}