Improved data visualisation through multiple dissimilarity modelling

Iain Rice

Research output: Contribution to journalArticle

Abstract

Popular dimension reduction and visualisation algorithms rely on the assumption that input dissimilarities are typically Euclidean, for instance Metric Multidimensional Scaling, t-distributed Stochastic Neighbour Embedding and the Gaussian Process Latent Variable Model. It is well known that this assumption does not hold for most datasets and often high-dimensional data sits upon a manifold of unknown global geometry. We present a method for improving the manifold charting process, coupled with Elastic MDS, such that we no longer assume that the manifold is Euclidean, or of any particular structure. We draw on the benefits of different dissimilarity measures allowing for the relative responsibilities, under a linear combination, to drive the visualisation process.
Original languageEnglish
Pages (from-to)288-302
Number of pages15
JournalInformation Sciences
Volume370-371
Early online date4 Aug 2016
DOIs
Publication statusPublished - 20 Nov 2016

Fingerprint

Data visualization
Data Visualization
Dissimilarity
Visualization
Euclidean
Modeling
Dissimilarity Measure
Latent Variable Models
Dimension Reduction
High-dimensional Data
Gaussian Process
Process Model
Geometry
Linear Combination
Scaling
Metric
Unknown
Multidimensional scaling
Latent variable models
Dimension reduction

Keywords

  • Dissimilarity
  • Euclidean
  • Multidimensional scaling
  • Visualisation

Cite this

@article{26cf82ec8b3d41b9baf635389cd308cd,
title = "Improved data visualisation through multiple dissimilarity modelling",
abstract = "Popular dimension reduction and visualisation algorithms rely on the assumption that input dissimilarities are typically Euclidean, for instance Metric Multidimensional Scaling, t-distributed Stochastic Neighbour Embedding and the Gaussian Process Latent Variable Model. It is well known that this assumption does not hold for most datasets and often high-dimensional data sits upon a manifold of unknown global geometry. We present a method for improving the manifold charting process, coupled with Elastic MDS, such that we no longer assume that the manifold is Euclidean, or of any particular structure. We draw on the benefits of different dissimilarity measures allowing for the relative responsibilities, under a linear combination, to drive the visualisation process.",
keywords = "Dissimilarity, Euclidean, Multidimensional scaling, Visualisation",
author = "Iain Rice",
year = "2016",
month = "11",
day = "20",
doi = "10.1016/j.ins.2016.07.073",
language = "English",
volume = "370-371",
pages = "288--302",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier",

}

Improved data visualisation through multiple dissimilarity modelling. / Rice, Iain.

In: Information Sciences, Vol. 370-371, 20.11.2016, p. 288-302.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Improved data visualisation through multiple dissimilarity modelling

AU - Rice, Iain

PY - 2016/11/20

Y1 - 2016/11/20

N2 - Popular dimension reduction and visualisation algorithms rely on the assumption that input dissimilarities are typically Euclidean, for instance Metric Multidimensional Scaling, t-distributed Stochastic Neighbour Embedding and the Gaussian Process Latent Variable Model. It is well known that this assumption does not hold for most datasets and often high-dimensional data sits upon a manifold of unknown global geometry. We present a method for improving the manifold charting process, coupled with Elastic MDS, such that we no longer assume that the manifold is Euclidean, or of any particular structure. We draw on the benefits of different dissimilarity measures allowing for the relative responsibilities, under a linear combination, to drive the visualisation process.

AB - Popular dimension reduction and visualisation algorithms rely on the assumption that input dissimilarities are typically Euclidean, for instance Metric Multidimensional Scaling, t-distributed Stochastic Neighbour Embedding and the Gaussian Process Latent Variable Model. It is well known that this assumption does not hold for most datasets and often high-dimensional data sits upon a manifold of unknown global geometry. We present a method for improving the manifold charting process, coupled with Elastic MDS, such that we no longer assume that the manifold is Euclidean, or of any particular structure. We draw on the benefits of different dissimilarity measures allowing for the relative responsibilities, under a linear combination, to drive the visualisation process.

KW - Dissimilarity

KW - Euclidean

KW - Multidimensional scaling

KW - Visualisation

UR - http://www.scopus.com/inward/record.url?scp=84982845643&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2016.07.073

DO - 10.1016/j.ins.2016.07.073

M3 - Article

VL - 370-371

SP - 288

EP - 302

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -