Files:DescriptionFile size FormatBrowse
Fulltext0.30 MBPDF (requires Acrobat Reader)Previous | Next
  
Authors:Stephen Ingram: University of British Columbia, Canada
Tamara Munzner: University of British Columbia, Canada
Publication title:Glint: An MDS Framework for Costly Distance Functions
Conference:Proceedings of SIGRAD 2012, Interactive Visual Analysis of Data, November 29–30, 2012, Växjö, Sweden
Publication type: Abstract and Fulltext
Issue:081
Article No.:005
Abstract:Previous algorithms for multidimensional scaling, or MDS, aim for scalable performance as the number of points to lay out increases. However, they either assume that the distance function is cheap to compute, and perform poorly when the distance function is costly, or they leave the precise number of distances to compute as a manual tuning parameter. We present Glint, an MDS algorithm framework that addresses both of these shortcomings. Glint is designed to automatically minimize the total number of distances computed by progressively computing a more and more densely sampled approximation of the distance matrix. We present instantiations of the Glint framework on three different classes of MDS algorithms: force-directed, analytic, and gradient-based. We validate the framework through computational benchmarks on several real-world datasets, and demonstrate substantial performance benefits without sacrificing layout quality.
Language:English
Keywords:I.3.3 [Human-centered Computing]: Visualization—Visualization systems and tools
Year:2012
No. of pages:10
Pages:29-38
ISBN:978-91-7519-723-4
Series:Linköping Electronic Conference Proceedings
ISSN (print):1650-3686
ISSN (online):1650-3740
File:http://www.ep.liu.se/ecp/081/005/ecp12081005.pdf
Available:2012-11-20
Publisher:Linköping University Electronic Press, Linköpings universitet

REFERENCE TO THIS PAGE
Stephen Ingram, Tamara Munzner (2012). Glint: An MDS Framework for Costly Distance Functions, Proceedings of SIGRAD 2012, Interactive Visual Analysis of Data, November 29–30, 2012, Växjö, Sweden http://www.ep.liu.se/ecp_article/index.en.aspx?issue=081;article=005 (accessed 4/16/2014)