Paper published in a journal (Scientific congresses, symposiums and conference proceedings)
GPU-Accelerated Mahalanobis-Average Hierarchical Clustering Analysis
Šmelko, Adam; KRATOCHVIL, Miroslav; Kruliš, Martin et al.
2021In Lecture Notes in Computer Science, 12820, p. 580-595
Peer reviewed
 

Files


Full Text
Šmelko2021_Chapter_GPU-AcceleratedMahalanobis-Ave.pdf
Publisher postprint (387.7 kB)
fulltext from Springer
Request a copy

All documents in ORBilu are protected by a user license.

Send to



Details



Keywords :
Clustering; High-dimensional data; GPU
Abstract :
[en] Hierarchical clustering is a common tool for simplification, exploration, and analysis of datasets in many areas of research. For data originating in flow cytometry, a specific variant of agglomerative clustering based Mahalanobis-average linkage has been shown to produce results better than the common linkages. However, the high complexity of computing the distance limits the applicability of the algorithm to datasets obtained from current equipment. We propose an optimized, GPU-accelerated open-source implementation of the Mahalanobis-average hierarchical clustering that improves the algorithm performance by over two orders of magnitude, thus allowing it to scale to the large datasets. We provide a detailed analysis of the optimizations and collected experimental results that are also portable to other hierarchical clustering algorithms; and demonstrate the use on realistic high-dimensional datasets.
Disciplines :
Computer science
Author, co-author :
Šmelko, Adam;  Charles University in Prague > Department of Software Engineering
KRATOCHVIL, Miroslav ;  University of Luxembourg > Luxembourg Centre for Systems Biomedicine (LCSB) > Bioinformatics Core
Kruliš, Martin;  Charles University in Prague > Department of Software Engineering
Sieger, Tomáš;  Czech Technical University in Prague > Department of Cybernetic
External co-authors :
yes
Language :
English
Title :
GPU-Accelerated Mahalanobis-Average Hierarchical Clustering Analysis
Publication date :
August 2021
Event name :
European Conference on Parallel Processing - Euro-Par 2021
Event date :
from 30-08-2021 to 3-09-2021
Audience :
International
Journal title :
Lecture Notes in Computer Science
ISSN :
0302-9743
eISSN :
1611-3349
Publisher :
Springer, Heidelberg, Germany
Volume :
12820
Pages :
580-595
Peer reviewed :
Peer reviewed
Focus Area :
Computational Sciences
Funders :
Czech Science Foundation (GAČR) project 19-22071Y
ELIXIR CZ LM2018131 (MEYS)
Charles University grant SVV-260451
Czech Health Research Council (AZV) [NV18-08-00385]
Available on ORBilu :
since 29 September 2021

Statistics


Number of views
102 (0 by Unilu)
Number of downloads
0 (0 by Unilu)

Scopus citations®
 
1
Scopus citations®
without self-citations
1
OpenCitations
 
0
OpenAlex citations
 
2
WoS citations
 
0

Bibliography


Similar publications



Contact ORBilu