Medoids in Almost-Linear Time via Multi-Armed Bandits
Vivek Bagaria, Govinda Kamath, Vasilis Ntranos, Martin Zhang, David Tse
Abstract
Computing the medoid of a large number of points in high-dimensional space is an increasingly common operation in many data science problems. We present an algorithm Med-dit to compute the medoid with high probability, which uses $O(n\log n)$ distance evaluations. Med-dit is based on a connection with the Multi-Armed Bandit problem. We evaluate the performance of Med-dit empirically on the Netflix-prize and single-cell RNA-seq datasets, containing hundreds of thousands of points living in tens of thousands of dimensions, and observe a $5$-$10$x improvement in performance over the current state of the art. We have released the code of Med-dit and our empirical results at https://github.com/bagavi/Meddit.
BibTeX
@InProceedings{pmlr-v84-bagaria18a,
title = {Medoids in Almost-Linear Time via Multi-Armed Bandits},
author = {Bagaria, Vivek and Kamath, Govinda and Ntranos, Vasilis and Zhang, Martin and Tse, David},
booktitle = {Proceedings of the Twenty-First International Conference on Artificial Intelligence and Statistics},
pages = {500--509},
year = {2018},
editor = {Storkey, Amos and Perez-Cruz, Fernando},
volume = {84},
series = {Proceedings of Machine Learning Research},
month = {09--11 Apr},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v84/bagaria18a/bagaria18a.pdf},
url = {https://proceedings.mlr.press/v84/bagaria18a.html},
abstract = {Computing the medoid of a large number of points in high-dimensional space is an increasingly common operation in many data science problems. We present an algorithm Med-dit to compute the medoid with high probability, which uses $O(n\log n)$ distance evaluations. Med-dit is based on a connection with the Multi-Armed Bandit problem. We evaluate the performance of Med-dit empirically on the Netflix-prize and single-cell RNA-seq datasets, containing hundreds of thousands of points living in tens of thousands of dimensions, and observe a $5$-$10$x improvement in performance over the current state of the art. We have released the code of Med-dit and our empirical results at https://github.com/bagavi/Meddit.}
}