Improving Deep Learning Projections by Neighborhood Analysis

Terri Modrakowski; Mateus Espadoto; Alexandre Falcao; Nina Hirata; Alex Telea

doi:10.1007/978-3-030-94893-1_6

Improving Deep Learning Projections by Neighborhood Analysis

Terri Modrakowski, Mateus Espadoto^*, Alexandre Falcao, Nina Hirata, Alex Telea

^*Corresponding author for this work

Utrecht University

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

Abstract

Visualization of multidimensional data is a difficult task, for which there are many tools. Among these tools, dimensionality reduction methods were shown to be particularly helpful to explore data visually. Techniques with good visual separation are very popular, such as those from the SNE-class, but those often are computationally expensive and non-parametric. An approach based on neural networks was recently proposed to address those shortcomings, but it introduces some fuzziness in the generated projection, which is not desired. In this paper we thoroughly explain the parameter space of this neural network approach and propose a new neighborhood-based learning paradigm, which further improves the quality of the projections learned by the neural networks, and we illustrate our approach on large real-world datasets.

Original language	English
Title of host publication	Computer Vision, Imaging and Computer Graphics Theory and Applications
Subtitle of host publication	15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers
Publisher	Springer
Pages	127–152
Number of pages	26
Edition	1
ISBN (Electronic)	978-3-030-94893-1
ISBN (Print)	978-3-030-94892-4
DOIs	https://doi.org/10.1007/978-3-030-94893-1_6
Publication status	Published - 23 Jan 2022

Publication series

Name	Communications in Computer and Information Science
Publisher	Springer
Volume	1474
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Keywords

Dimensionality reduction
Machine learning
Neural networks
Multidimensional projections

Access to Document

10.1007/978-3-030-94893-1_6

978-3-030-94893-1_6Final published version, 17.8 MBLicence: Taverne

Cite this

Modrakowski, T., Espadoto, M., Falcao, A., Hirata, N., & Telea, A. (2022). Improving Deep Learning Projections by Neighborhood Analysis. In Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers (1 ed., pp. 127–152). (Communications in Computer and Information Science; Vol. 1474). Springer. https://doi.org/10.1007/978-3-030-94893-1_6

Modrakowski, Terri ; Espadoto, Mateus ; Falcao, Alexandre et al. / Improving Deep Learning Projections by Neighborhood Analysis. Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers. 1. ed. Springer, 2022. pp. 127–152 (Communications in Computer and Information Science).

@inbook{5c10ee0c7896453ab456626a9400ef00,

title = "Improving Deep Learning Projections by Neighborhood Analysis",

abstract = "Visualization of multidimensional data is a difficult task, for which there are many tools. Among these tools, dimensionality reduction methods were shown to be particularly helpful to explore data visually. Techniques with good visual separation are very popular, such as those from the SNE-class, but those often are computationally expensive and non-parametric. An approach based on neural networks was recently proposed to address those shortcomings, but it introduces some fuzziness in the generated projection, which is not desired. In this paper we thoroughly explain the parameter space of this neural network approach and propose a new neighborhood-based learning paradigm, which further improves the quality of the projections learned by the neural networks, and we illustrate our approach on large real-world datasets.",

keywords = "Dimensionality reduction, Machine learning, Neural networks, Multidimensional projections",

author = "Terri Modrakowski and Mateus Espadoto and Alexandre Falcao and Nina Hirata and Alex Telea",

year = "2022",

month = jan,

day = "23",

doi = "10.1007/978-3-030-94893-1_6",

language = "English",

isbn = "978-3-030-94892-4",

series = "Communications in Computer and Information Science",

publisher = "Springer",

pages = "127–152",

booktitle = "Computer Vision, Imaging and Computer Graphics Theory and Applications",

address = "Germany",

edition = "1",

}

Modrakowski, T, Espadoto, M, Falcao, A, Hirata, N & Telea, A 2022, Improving Deep Learning Projections by Neighborhood Analysis. in Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers. 1 edn, Communications in Computer and Information Science, vol. 1474, Springer, pp. 127–152. https://doi.org/10.1007/978-3-030-94893-1_6

Improving Deep Learning Projections by Neighborhood Analysis. / Modrakowski, Terri; Espadoto, Mateus; Falcao, Alexandre et al.
Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers. 1. ed. Springer, 2022. p. 127–152 (Communications in Computer and Information Science; Vol. 1474).

Research output: Chapter in Book/Report/Conference proceeding › Chapter › Academic › peer-review

TY - CHAP

T1 - Improving Deep Learning Projections by Neighborhood Analysis

AU - Modrakowski, Terri

AU - Espadoto, Mateus

AU - Falcao, Alexandre

AU - Hirata, Nina

AU - Telea, Alex

PY - 2022/1/23

Y1 - 2022/1/23

N2 - Visualization of multidimensional data is a difficult task, for which there are many tools. Among these tools, dimensionality reduction methods were shown to be particularly helpful to explore data visually. Techniques with good visual separation are very popular, such as those from the SNE-class, but those often are computationally expensive and non-parametric. An approach based on neural networks was recently proposed to address those shortcomings, but it introduces some fuzziness in the generated projection, which is not desired. In this paper we thoroughly explain the parameter space of this neural network approach and propose a new neighborhood-based learning paradigm, which further improves the quality of the projections learned by the neural networks, and we illustrate our approach on large real-world datasets.

AB - Visualization of multidimensional data is a difficult task, for which there are many tools. Among these tools, dimensionality reduction methods were shown to be particularly helpful to explore data visually. Techniques with good visual separation are very popular, such as those from the SNE-class, but those often are computationally expensive and non-parametric. An approach based on neural networks was recently proposed to address those shortcomings, but it introduces some fuzziness in the generated projection, which is not desired. In this paper we thoroughly explain the parameter space of this neural network approach and propose a new neighborhood-based learning paradigm, which further improves the quality of the projections learned by the neural networks, and we illustrate our approach on large real-world datasets.

KW - Dimensionality reduction

KW - Machine learning

KW - Neural networks

KW - Multidimensional projections

U2 - 10.1007/978-3-030-94893-1_6

DO - 10.1007/978-3-030-94893-1_6

M3 - Chapter

SN - 978-3-030-94892-4

T3 - Communications in Computer and Information Science

SP - 127

EP - 152

BT - Computer Vision, Imaging and Computer Graphics Theory and Applications

PB - Springer

ER -

Modrakowski T, Espadoto M, Falcao A, Hirata N, Telea A. Improving Deep Learning Projections by Neighborhood Analysis. In Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers. 1 ed. Springer. 2022. p. 127–152. (Communications in Computer and Information Science). doi: 10.1007/978-3-030-94893-1_6

Improving Deep Learning Projections by Neighborhood Analysis

Abstract

Publication series

Keywords

Access to Document

Fingerprint

Cite this