Horizontally Scalable Submodular Maximization

Mario Lucic, Olivier Bachem, Morteza Zadimoghaddam, Andreas Krause
Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:2981-2989, 2016.

Abstract

A variety of large-scale machine learning problems can be cast as instances of constrained submodular maximization. Existing approaches for distributed submodular maximization have a critical drawback: The capacity - number of instances that can fit in memory - must grow with the data set size. In practice, while one can provision many machines, the capacity of each machine is limited by physical constraints. We propose a truly scalable approach for distributed submodular maximization under fixed capacity. The proposed framework applies to a broad class of algorithms and constraints and provides theoretical guarantees on the approximation factor for any available capacity. We empirically evaluate the proposed algorithm on a variety of data sets and demonstrate that it achieves performance competitive with the centralized greedy solution.

Cite this Paper

BibTeX
@InProceedings{pmlr-v48-lucic16, title = {Horizontally Scalable Submodular Maximization}, author = {Lucic, Mario and Bachem, Olivier and Zadimoghaddam, Morteza and Krause, Andreas}, booktitle = {Proceedings of The 33rd International Conference on Machine Learning}, pages = {2981--2989}, 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/lucic16.pdf}, url = {https://proceedings.mlr.press/v48/lucic16.html}, abstract = {A variety of large-scale machine learning problems can be cast as instances of constrained submodular maximization. Existing approaches for distributed submodular maximization have a critical drawback: The capacity - number of instances that can fit in memory - must grow with the data set size. In practice, while one can provision many machines, the capacity of each machine is limited by physical constraints. We propose a truly scalable approach for distributed submodular maximization under fixed capacity. The proposed framework applies to a broad class of algorithms and constraints and provides theoretical guarantees on the approximation factor for any available capacity. We empirically evaluate the proposed algorithm on a variety of data sets and demonstrate that it achieves performance competitive with the centralized greedy solution.} }
Endnote
%0 Conference Paper %T Horizontally Scalable Submodular Maximization %A Mario Lucic %A Olivier Bachem %A Morteza Zadimoghaddam %A Andreas Krause %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-lucic16 %I PMLR %P 2981--2989 %U https://proceedings.mlr.press/v48/lucic16.html %V 48 %X A variety of large-scale machine learning problems can be cast as instances of constrained submodular maximization. Existing approaches for distributed submodular maximization have a critical drawback: The capacity - number of instances that can fit in memory - must grow with the data set size. In practice, while one can provision many machines, the capacity of each machine is limited by physical constraints. We propose a truly scalable approach for distributed submodular maximization under fixed capacity. The proposed framework applies to a broad class of algorithms and constraints and provides theoretical guarantees on the approximation factor for any available capacity. We empirically evaluate the proposed algorithm on a variety of data sets and demonstrate that it achieves performance competitive with the centralized greedy solution.
RIS
TY - CPAPER TI - Horizontally Scalable Submodular Maximization AU - Mario Lucic AU - Olivier Bachem AU - Morteza Zadimoghaddam AU - Andreas Krause 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-lucic16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 48 SP - 2981 EP - 2989 L1 - http://proceedings.mlr.press/v48/lucic16.pdf UR - https://proceedings.mlr.press/v48/lucic16.html AB - A variety of large-scale machine learning problems can be cast as instances of constrained submodular maximization. Existing approaches for distributed submodular maximization have a critical drawback: The capacity - number of instances that can fit in memory - must grow with the data set size. In practice, while one can provision many machines, the capacity of each machine is limited by physical constraints. We propose a truly scalable approach for distributed submodular maximization under fixed capacity. The proposed framework applies to a broad class of algorithms and constraints and provides theoretical guarantees on the approximation factor for any available capacity. We empirically evaluate the proposed algorithm on a variety of data sets and demonstrate that it achieves performance competitive with the centralized greedy solution. ER -
APA
Lucic, M., Bachem, O., Zadimoghaddam, M. & Krause, A.. (2016). Horizontally Scalable Submodular Maximization. Proceedings of The 33rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 48:2981-2989 Available from https://proceedings.mlr.press/v48/lucic16.html.

Related Material