Distributed Mean Estimation with Limited Communication

Ananda Theertha Suresh, Felix X. Yu, Sanjiv Kumar, H. Brendan McMahan
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:3329-3337, 2017.

Abstract

Motivated by the need for distributed learning and optimization algorithms with low communication cost, we study communication efficient algorithms for distributed mean estimation. Unlike previous works, we make no probabilistic assumptions on the data. We first show that for $d$ dimensional data with $n$ clients, a naive stochastic rounding approach yields a mean squared error (MSE) of $\Theta(d/n)$ and uses a constant number of bits per dimension per client. We then extend this naive algorithm in two ways: we show that applying a structured random rotation before quantization reduces the error to $\mathcal{O}((\log d)/n)$ and a better coding strategy further reduces the error to $\mathcal{O}(1/n)$. We also show that the latter coding strategy is optimal up to a constant in the minimax sense i.e., it achieves the best MSE for a given communication cost. We finally demonstrate the practicality of our algorithms by applying them to distributed Lloyd’s algorithm for k-means and power iteration for PCA.

Cite this Paper


BibTeX
@InProceedings{pmlr-v70-suresh17a, title = {Distributed Mean Estimation with Limited Communication}, author = {Ananda Theertha Suresh and Felix X. Yu and Sanjiv Kumar and H. Brendan McMahan}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {3329--3337}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/suresh17a/suresh17a.pdf}, url = {https://proceedings.mlr.press/v70/suresh17a.html}, abstract = {Motivated by the need for distributed learning and optimization algorithms with low communication cost, we study communication efficient algorithms for distributed mean estimation. Unlike previous works, we make no probabilistic assumptions on the data. We first show that for $d$ dimensional data with $n$ clients, a naive stochastic rounding approach yields a mean squared error (MSE) of $\Theta(d/n)$ and uses a constant number of bits per dimension per client. We then extend this naive algorithm in two ways: we show that applying a structured random rotation before quantization reduces the error to $\mathcal{O}((\log d)/n)$ and a better coding strategy further reduces the error to $\mathcal{O}(1/n)$. We also show that the latter coding strategy is optimal up to a constant in the minimax sense i.e., it achieves the best MSE for a given communication cost. We finally demonstrate the practicality of our algorithms by applying them to distributed Lloyd’s algorithm for k-means and power iteration for PCA.} }
Endnote
%0 Conference Paper %T Distributed Mean Estimation with Limited Communication %A Ananda Theertha Suresh %A Felix X. Yu %A Sanjiv Kumar %A H. Brendan McMahan %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-suresh17a %I PMLR %P 3329--3337 %U https://proceedings.mlr.press/v70/suresh17a.html %V 70 %X Motivated by the need for distributed learning and optimization algorithms with low communication cost, we study communication efficient algorithms for distributed mean estimation. Unlike previous works, we make no probabilistic assumptions on the data. We first show that for $d$ dimensional data with $n$ clients, a naive stochastic rounding approach yields a mean squared error (MSE) of $\Theta(d/n)$ and uses a constant number of bits per dimension per client. We then extend this naive algorithm in two ways: we show that applying a structured random rotation before quantization reduces the error to $\mathcal{O}((\log d)/n)$ and a better coding strategy further reduces the error to $\mathcal{O}(1/n)$. We also show that the latter coding strategy is optimal up to a constant in the minimax sense i.e., it achieves the best MSE for a given communication cost. We finally demonstrate the practicality of our algorithms by applying them to distributed Lloyd’s algorithm for k-means and power iteration for PCA.
APA
Suresh, A.T., Yu, F.X., Kumar, S. & McMahan, H.B.. (2017). Distributed Mean Estimation with Limited Communication. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:3329-3337 Available from https://proceedings.mlr.press/v70/suresh17a.html.

Related Material