Fully Dynamic Algorithm for Constrained Submodular Optimization
Abstract
The task of maximizing a monotone submodular function under a cardinality constraint is at the core of many machine learning and data mining applications, including data summarization, sparse regression and coverage problems. We study this classic problem in the fully dynamic setting, where elements can be both inserted and removed. Our main result is a randomized algorithm that maintains an efficient data structure with a poly-logarithmic amortized update time and yields a $(1/2-epsilon)$-approximate solution. We complement our theoretical analysis with an empirical study of the performance of our algorithm.
Cite
Text
Lattanzi et al. "Fully Dynamic Algorithm for Constrained Submodular Optimization." Neural Information Processing Systems, 2020.Markdown
[Lattanzi et al. "Fully Dynamic Algorithm for Constrained Submodular Optimization." Neural Information Processing Systems, 2020.](https://mlanthology.org/neurips/2020/lattanzi2020neurips-fully/)BibTeX
@inproceedings{lattanzi2020neurips-fully,
title = {{Fully Dynamic Algorithm for Constrained Submodular Optimization}},
author = {Lattanzi, Silvio and Mitrović, Slobodan and Norouzi-Fard, Ashkan and Tarnawski, Jakub M and Zadimoghaddam, Morteza},
booktitle = {Neural Information Processing Systems},
year = {2020},
url = {https://mlanthology.org/neurips/2020/lattanzi2020neurips-fully/}
}