Parallel and Distributed Block-Coordinate Frank-Wolfe Algorithms

Yu-Xiang Wang, Veeranjaneyulu Sadhanala, Wei Dai, Willie Neiswanger, Suvrit Sra, Eric Xing
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:1548-1557, 2016.

Abstract

We study parallel and distributed Frank-Wolfe algorithms; the former on shared memory machines with mini-batching, and the latter in a delayed update framework. In both cases, we perform computations asynchronously whenever possible. We assume block-separable constraints as in Block-Coordinate Frank-Wolfe (BCFW) method (Lacoste et. al., 2013) , but our analysis subsumes BCFW and reveals problem-dependent quantities that govern the speedups of our methods over BCFW. A notable feature of our algorithms is that they do not depend on worst-case bounded delays, but only (mildly) on **expected** delays, making them robust to stragglers and faulty worker threads. We present experiments on structural SVM and Group Fused Lasso, and observe significant speedups over competing state-of-the-art (and synchronous) methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-wangd16, title = {Parallel and Distributed Block-Coordinate Frank-Wolfe Algorithms}, author = {Wang, Yu-Xiang and Sadhanala, Veeranjaneyulu and Dai, Wei and Neiswanger, Willie and Sra, Suvrit and Xing, Eric}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {1548--1557}, year = {2016}, editor = {Balcan, Maria Florina and Weinberger, Kilian Q.}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/wangd16.pdf}, url = {https://proceedings.mlr.press/v48/wangd16.html}, abstract = {We study parallel and distributed Frank-Wolfe algorithms; the former on shared memory machines with mini-batching, and the latter in a delayed update framework. In both cases, we perform computations asynchronously whenever possible. We assume block-separable constraints as in Block-Coordinate Frank-Wolfe (BCFW) method (Lacoste et. al., 2013) , but our analysis subsumes BCFW and reveals problem-dependent quantities that govern the speedups of our methods over BCFW. A notable feature of our algorithms is that they do not depend on worst-case bounded delays, but only (mildly) on **expected** delays, making them robust to stragglers and faulty worker threads. We present experiments on structural SVM and Group Fused Lasso, and observe significant speedups over competing state-of-the-art (and synchronous) methods.} }
Endnote
%0 Conference Paper %T Parallel and Distributed Block-Coordinate Frank-Wolfe Algorithms %A Yu-Xiang Wang %A Veeranjaneyulu Sadhanala %A Wei Dai %A Willie Neiswanger %A Suvrit Sra %A Eric Xing %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-wangd16 %I PMLR %P 1548--1557 %U https://proceedings.mlr.press/v48/wangd16.html %V 48 %X We study parallel and distributed Frank-Wolfe algorithms; the former on shared memory machines with mini-batching, and the latter in a delayed update framework. In both cases, we perform computations asynchronously whenever possible. We assume block-separable constraints as in Block-Coordinate Frank-Wolfe (BCFW) method (Lacoste et. al., 2013) , but our analysis subsumes BCFW and reveals problem-dependent quantities that govern the speedups of our methods over BCFW. A notable feature of our algorithms is that they do not depend on worst-case bounded delays, but only (mildly) on **expected** delays, making them robust to stragglers and faulty worker threads. We present experiments on structural SVM and Group Fused Lasso, and observe significant speedups over competing state-of-the-art (and synchronous) methods.
RIS
TY - CPAPER TI - Parallel and Distributed Block-Coordinate Frank-Wolfe Algorithms AU - Yu-Xiang Wang AU - Veeranjaneyulu Sadhanala AU - Wei Dai AU - Willie Neiswanger AU - Suvrit Sra AU - Eric Xing BT - Proceedings of The 33rd International Conference on Machine Learning DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-wangd16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 1548 EP - 1557 L1 - http://proceedings.mlr.press/v48/wangd16.pdf UR - https://proceedings.mlr.press/v48/wangd16.html AB - We study parallel and distributed Frank-Wolfe algorithms; the former on shared memory machines with mini-batching, and the latter in a delayed update framework. In both cases, we perform computations asynchronously whenever possible. We assume block-separable constraints as in Block-Coordinate Frank-Wolfe (BCFW) method (Lacoste et. al., 2013) , but our analysis subsumes BCFW and reveals problem-dependent quantities that govern the speedups of our methods over BCFW. A notable feature of our algorithms is that they do not depend on worst-case bounded delays, but only (mildly) on **expected** delays, making them robust to stragglers and faulty worker threads. We present experiments on structural SVM and Group Fused Lasso, and observe significant speedups over competing state-of-the-art (and synchronous) methods. ER -
APA
Wang, Y., Sadhanala, V., Dai, W., Neiswanger, W., Sra, S. & Xing, E.. (2016). Parallel and Distributed Block-Coordinate Frank-Wolfe Algorithms. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:1548-1557 Available from https://proceedings.mlr.press/v48/wangd16.html.

Related Material