Parallel k-means++ for Multiple Shared-Memory Architectures

Conference · Thu Sep 22 00:00:00 EDT 2016

DOI:https://doi.org/10.1109/ICPP.2016.18· OSTI ID:1334876

Mackey, Patrick S.; Lewis, Robert R.

In recent years k-means++ has become a popular initialization technique for improved k-means clustering. To date, most of the work done to improve its performance has involved parallelizing algorithms that are only approximations of k-means++. In this paper we present a parallelization of the exact k-means++ algorithm, with a proof of its correctness. We develop implementations for three distinct shared-memory architectures: multicore CPU, high performance GPU, and the massively multithreaded Cray XMT platform. We demonstrate the scalability of the algorithm on each platform. In addition we present a visual approach for showing which platform performed k-means++ the fastest for varying data sizes.

• Research Organization: Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1334876
• Report Number(s): PNNL-SA-116273
• Resource Relation: Conference: 45th International Conference on Parallel Processing (ICPP 2016), August 16-19, 2016, Philadelphia, PA, 93-102
• Country of Publication: United States
• Language: English

Related Subjects

data clustering
GPGPU
multithreaded architecture
OpenMP