Semi-supervised learning of hierarchical latent trait models for data visualisation

Yi Sun, Peter Tino, Ata Kaban, Ian T. Nabney

Research output: Working paperTechnical report

Abstract

An interactive hierarchical Generative Topographic Mapping (HGTM) ¸iteH<sub>G</sub>TM has been developed to visualise complex data sets. In this paper, we build a more general visualisation system by extending the HGTM visualisation system in 3 directions: bf (1) We generalize HGTM to noise models from the exponential family of distributions. The basic building block is the Latent Trait Model (LTM) developed in ¸iteKaban<sub>p</sub>ami. bf (2) We give the user a choice of initializing the child plots of the current plot in either em interactive, or em automatic mode. In the interactive mode the user interactively selects ``regions of interest'' as in ¸iteH<sub>G</sub>TM, whereas in the automatic mode an unsupervised minimum message length (MML)-driven construction of a mixture of LTMs is employed. bf (3) We derive general formulas for magnification factors in latent trait models. Magnification factors are a useful tool to improve our understanding of the visualisation plots, since they can highlight the boundaries between data clusters. The unsupervised construction is particularly useful when high-level plots are covered with dense clusters of highly overlapping data projections, making it difficult to use the interactive mode. Such a situation often arises when visualizing large data sets. We illustrate our approach on a toy example and apply our system to three more complex real data sets.
Original languageEnglish
Place of PublicationBirmingham, UK
PublisherAston University
Number of pages27
Publication statusPublished - 2002

Fingerprint

Data visualization
Supervised learning
Visualization

Keywords

  • hierarchical Generative Topographic Mapping
  • interactive mode
  • automatic mode
  • magnification factors
  • latent trait models
  • unsupervised construction
  • overlapping data

Cite this

Sun, Y., Tino, P., Kaban, A., & Nabney, I. T. (2002). Semi-supervised learning of hierarchical latent trait models for data visualisation. Birmingham, UK: Aston University.
Sun, Yi ; Tino, Peter ; Kaban, Ata ; Nabney, Ian T. / Semi-supervised learning of hierarchical latent trait models for data visualisation. Birmingham, UK : Aston University, 2002.
@techreport{251307b82193472f91e0184c6360144c,
title = "Semi-supervised learning of hierarchical latent trait models for data visualisation",
abstract = "An interactive hierarchical Generative Topographic Mapping (HGTM) ¸iteHGTM has been developed to visualise complex data sets. In this paper, we build a more general visualisation system by extending the HGTM visualisation system in 3 directions: bf (1) We generalize HGTM to noise models from the exponential family of distributions. The basic building block is the Latent Trait Model (LTM) developed in ¸iteKabanpami. bf (2) We give the user a choice of initializing the child plots of the current plot in either em interactive, or em automatic mode. In the interactive mode the user interactively selects ``regions of interest'' as in ¸iteHGTM, whereas in the automatic mode an unsupervised minimum message length (MML)-driven construction of a mixture of LTMs is employed. bf (3) We derive general formulas for magnification factors in latent trait models. Magnification factors are a useful tool to improve our understanding of the visualisation plots, since they can highlight the boundaries between data clusters. The unsupervised construction is particularly useful when high-level plots are covered with dense clusters of highly overlapping data projections, making it difficult to use the interactive mode. Such a situation often arises when visualizing large data sets. We illustrate our approach on a toy example and apply our system to three more complex real data sets.",
keywords = "hierarchical Generative Topographic Mapping, interactive mode, automatic mode, magnification factors, latent trait models, unsupervised construction, overlapping data",
author = "Yi Sun and Peter Tino and Ata Kaban and Nabney, {Ian T.}",
year = "2002",
language = "English",
publisher = "Aston University",
type = "WorkingPaper",
institution = "Aston University",

}

Sun, Y, Tino, P, Kaban, A & Nabney, IT 2002 'Semi-supervised learning of hierarchical latent trait models for data visualisation' Aston University, Birmingham, UK.

Semi-supervised learning of hierarchical latent trait models for data visualisation. / Sun, Yi; Tino, Peter; Kaban, Ata; Nabney, Ian T.

Birmingham, UK : Aston University, 2002.

Research output: Working paperTechnical report

TY - UNPB

T1 - Semi-supervised learning of hierarchical latent trait models for data visualisation

AU - Sun, Yi

AU - Tino, Peter

AU - Kaban, Ata

AU - Nabney, Ian T.

PY - 2002

Y1 - 2002

N2 - An interactive hierarchical Generative Topographic Mapping (HGTM) ¸iteHGTM has been developed to visualise complex data sets. In this paper, we build a more general visualisation system by extending the HGTM visualisation system in 3 directions: bf (1) We generalize HGTM to noise models from the exponential family of distributions. The basic building block is the Latent Trait Model (LTM) developed in ¸iteKabanpami. bf (2) We give the user a choice of initializing the child plots of the current plot in either em interactive, or em automatic mode. In the interactive mode the user interactively selects ``regions of interest'' as in ¸iteHGTM, whereas in the automatic mode an unsupervised minimum message length (MML)-driven construction of a mixture of LTMs is employed. bf (3) We derive general formulas for magnification factors in latent trait models. Magnification factors are a useful tool to improve our understanding of the visualisation plots, since they can highlight the boundaries between data clusters. The unsupervised construction is particularly useful when high-level plots are covered with dense clusters of highly overlapping data projections, making it difficult to use the interactive mode. Such a situation often arises when visualizing large data sets. We illustrate our approach on a toy example and apply our system to three more complex real data sets.

AB - An interactive hierarchical Generative Topographic Mapping (HGTM) ¸iteHGTM has been developed to visualise complex data sets. In this paper, we build a more general visualisation system by extending the HGTM visualisation system in 3 directions: bf (1) We generalize HGTM to noise models from the exponential family of distributions. The basic building block is the Latent Trait Model (LTM) developed in ¸iteKabanpami. bf (2) We give the user a choice of initializing the child plots of the current plot in either em interactive, or em automatic mode. In the interactive mode the user interactively selects ``regions of interest'' as in ¸iteHGTM, whereas in the automatic mode an unsupervised minimum message length (MML)-driven construction of a mixture of LTMs is employed. bf (3) We derive general formulas for magnification factors in latent trait models. Magnification factors are a useful tool to improve our understanding of the visualisation plots, since they can highlight the boundaries between data clusters. The unsupervised construction is particularly useful when high-level plots are covered with dense clusters of highly overlapping data projections, making it difficult to use the interactive mode. Such a situation often arises when visualizing large data sets. We illustrate our approach on a toy example and apply our system to three more complex real data sets.

KW - hierarchical Generative Topographic Mapping

KW - interactive mode

KW - automatic mode

KW - magnification factors

KW - latent trait models

KW - unsupervised construction

KW - overlapping data

M3 - Technical report

BT - Semi-supervised learning of hierarchical latent trait models for data visualisation

PB - Aston University

CY - Birmingham, UK

ER -

Sun Y, Tino P, Kaban A, Nabney IT. Semi-supervised learning of hierarchical latent trait models for data visualisation. Birmingham, UK: Aston University. 2002.