Supervised dimensionality reduction technique accounting for soft classes

Sorina Mustatea; Michaël Aupetit; Jaakko Peltonen; Sylvain Lespinats; Denys Dutykh

doi:10.14428/esann/2022.ES2022-26

Supervised dimensionality reduction technique accounting for soft classes

Sorina Mustatea
, Michaël Aupetit
, Jaakko Peltonen
, Sylvain Lespinats
, Denys Dutykh

Qatar Computing Research Institute

Université Grenoble Alpes
Tampere University

Research output: Contribution to conference › Paper › peer-review

Abstract

Exploratory visual analysis of multidimensional labeled data is challenging. Multidimensional Projections for labeled data attempt to separate classes while preserving neighborhoods. In this work, we consider the case where instances are assigned multiple labels with probabilities or weights: for example, the output of a probabilistic classifier, fuzzy membership functions in fuzzy logic, or the share of votes for each candidate in an election. We propose a new technique to better preserve neighborhoods of such data. Our experiments show improved qualitative results compared to unsupervised, and existing dimensionality reduction techniques.

Original language	English
Pages	13-18
Number of pages	6
DOIs	https://doi.org/10.14428/esann/2022.ES2022-26
Publication status	Published - 2022
Event	30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022 - Bruges, Belgium Duration: 5 Oct 2022 → 7 Oct 2022

Conference

Conference	30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022
Country/Territory	Belgium
City	Bruges
Period	5/10/22 → 7/10/22

Access to Document

10.14428/esann/2022.ES2022-26

Cite this

@conference{4720e68811e34d4c955b0702cb9c42c0,

title = "Supervised dimensionality reduction technique accounting for soft classes",

abstract = "Exploratory visual analysis of multidimensional labeled data is challenging. Multidimensional Projections for labeled data attempt to separate classes while preserving neighborhoods. In this work, we consider the case where instances are assigned multiple labels with probabilities or weights: for example, the output of a probabilistic classifier, fuzzy membership functions in fuzzy logic, or the share of votes for each candidate in an election. We propose a new technique to better preserve neighborhoods of such data. Our experiments show improved qualitative results compared to unsupervised, and existing dimensionality reduction techniques.",

author = "Sorina Mustatea and Micha{\"e}l Aupetit and Jaakko Peltonen and Sylvain Lespinats and Denys Dutykh",

note = "Publisher Copyright: {\textcopyright} 2022 ESANN 2022 Proceedings - 30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. All rights reserved.; 30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022 ; Conference date: 05-10-2022 Through 07-10-2022",

year = "2022",

doi = "10.14428/esann/2022.ES2022-26",

language = "English",

pages = "13--18",

}

Mustatea, S, Aupetit, M, Peltonen, J, Lespinats, S & Dutykh, D 2022, 'Supervised dimensionality reduction technique accounting for soft classes', Paper presented at 30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022, Bruges, Belgium, 5/10/22 - 7/10/22 pp. 13-18. https://doi.org/10.14428/esann/2022.ES2022-26

Supervised dimensionality reduction technique accounting for soft classes. / Mustatea, Sorina; Aupetit, Michaël; Peltonen, Jaakko et al.
2022. 13-18 Paper presented at 30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022, Bruges, Belgium.

Research output: Contribution to conference › Paper › peer-review

TY - CONF

T1 - Supervised dimensionality reduction technique accounting for soft classes

AU - Mustatea, Sorina

AU - Aupetit, Michaël

AU - Peltonen, Jaakko

AU - Lespinats, Sylvain

AU - Dutykh, Denys

PY - 2022

Y1 - 2022

N2 - Exploratory visual analysis of multidimensional labeled data is challenging. Multidimensional Projections for labeled data attempt to separate classes while preserving neighborhoods. In this work, we consider the case where instances are assigned multiple labels with probabilities or weights: for example, the output of a probabilistic classifier, fuzzy membership functions in fuzzy logic, or the share of votes for each candidate in an election. We propose a new technique to better preserve neighborhoods of such data. Our experiments show improved qualitative results compared to unsupervised, and existing dimensionality reduction techniques.

AB - Exploratory visual analysis of multidimensional labeled data is challenging. Multidimensional Projections for labeled data attempt to separate classes while preserving neighborhoods. In this work, we consider the case where instances are assigned multiple labels with probabilities or weights: for example, the output of a probabilistic classifier, fuzzy membership functions in fuzzy logic, or the share of votes for each candidate in an election. We propose a new technique to better preserve neighborhoods of such data. Our experiments show improved qualitative results compared to unsupervised, and existing dimensionality reduction techniques.

UR - https://www.scopus.com/pages/publications/105041765330

U2 - 10.14428/esann/2022.ES2022-26

DO - 10.14428/esann/2022.ES2022-26

M3 - Paper

AN - SCOPUS:105041765330

SP - 13

EP - 18

T2 - 30th European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2022

Y2 - 5 October 2022 through 7 October 2022

ER -

Supervised dimensionality reduction technique accounting for soft classes

Abstract

Conference

Access to Document

Other files and links

Fingerprint

Cite this