Evaluation of a new representation for noise reduction in distant supervision

Juan Luis García-Mendoza; Luis Villaseñor-Pineda; Davide Buscaldi; Lázaro Bustio-Martínez; Felipe Orihuela-Espina

doi:10.1007/978-3-031-19496-2_8

Evaluation of a new representation for noise reduction in distant supervision

Juan Luis García-Mendoza^*, Luis Villaseñor-Pineda, Davide Buscaldi, Lázaro Bustio-Martínez, Felipe Orihuela-Espina

^*Corresponding author for this work

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

13 Downloads (Pure)

Abstract

Distant Supervision is a relation extraction approach that allows automatic labeling of a dataset. However, this labeling introduces noise in the labels (e.g., when two entities in a sentence are automatically labeled with an invalid relation). Noise in labels makes difficult the relation extraction task. This noise is precisely one of the main challenges of this task. Until now, the methods that incorporate a previous noise reduction step do not evaluate the performance of this step. This paper evaluates the noise reduction using a new representation obtained with autoencoders. In addition, it was incoporated more information to the input of the autoencoder proposed in the state-of-the-art to improve the representation over which the noise is reduced. Also, three methods were proposed to select the instances considered as real. As a result, it was obtained the highest values of the area under the ROC curves using the improved input combined with state-of-the-art anomaly detection methods. Moreover, the three proposed selection methods significantly improve the existing method in the literature.

Original language	English
Title of host publication	Advances in Computational Intelligence
Subtitle of host publication	21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II
Editors	Obdulia Pichardo Lagunas, Bella Martínez Seis, Juan Martínez-Miranda
Publisher	Springer
Pages	101-113
Number of pages	13
Edition	1
ISBN (Electronic)	9783031194962
ISBN (Print)	9783031194955
DOIs	https://doi.org/10.1007/978-3-031-19496-2_8
Publication status	Published - 25 Oct 2022
Event	21st Mexican International Conference on Artificial Intelligence, MICAI 2022 - Monterrey, Mexico Duration: 24 Oct 2022 → 29 Oct 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13613 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	21st Mexican International Conference on Artificial Intelligence, MICAI 2022
Country/Territory	Mexico
City	Monterrey
Period	24/10/22 → 29/10/22

Bibliographical note

Funding Information:
The present work was supported by CONACyT/México (schol-arship 937210 and grant CB-2015-01-257383) and Labex EFL through EFL mobility grants. Additionally, the authors thank CONACYT for the computer resources provided through the INAOE Supercomputing Laboratory’s Deep Learning Platform for Language Technologies.

Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Keywords

Adversarial autoencoders
Data representation
Distant supervision
Noise reduction

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-031-19496-2_8

GarcíaMendozaJ2022Evaluation
This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1007/978-3-031-19496-2_8
Accepted author manuscript, 598 KBLicence: Other (please specify with Rights Statement)

Cite this

García-Mendoza, J. L., Villaseñor-Pineda, L., Buscaldi, D., Bustio-Martínez, L., & Orihuela-Espina, F. (2022). Evaluation of a new representation for noise reduction in distant supervision. In O. Pichardo Lagunas, B. Martínez Seis, & J. Martínez-Miranda (Eds.), Advances in Computational Intelligence: 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II (1 ed., pp. 101-113). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13613 LNAI). Springer. https://doi.org/10.1007/978-3-031-19496-2_8

García-Mendoza, Juan Luis ; Villaseñor-Pineda, Luis ; Buscaldi, Davide et al. / Evaluation of a new representation for noise reduction in distant supervision. Advances in Computational Intelligence: 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II. editor / Obdulia Pichardo Lagunas ; Bella Martínez Seis ; Juan Martínez-Miranda. 1. ed. Springer, 2022. pp. 101-113 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{f36dc814f811442cb24255c4a0443cac,

title = "Evaluation of a new representation for noise reduction in distant supervision",

abstract = "Distant Supervision is a relation extraction approach that allows automatic labeling of a dataset. However, this labeling introduces noise in the labels (e.g., when two entities in a sentence are automatically labeled with an invalid relation). Noise in labels makes difficult the relation extraction task. This noise is precisely one of the main challenges of this task. Until now, the methods that incorporate a previous noise reduction step do not evaluate the performance of this step. This paper evaluates the noise reduction using a new representation obtained with autoencoders. In addition, it was incoporated more information to the input of the autoencoder proposed in the state-of-the-art to improve the representation over which the noise is reduced. Also, three methods were proposed to select the instances considered as real. As a result, it was obtained the highest values of the area under the ROC curves using the improved input combined with state-of-the-art anomaly detection methods. Moreover, the three proposed selection methods significantly improve the existing method in the literature.",

keywords = "Adversarial autoencoders, Data representation, Distant supervision, Noise reduction",

author = "Garc{\'i}a-Mendoza, {Juan Luis} and Luis Villase{\~n}or-Pineda and Davide Buscaldi and L{\'a}zaro Bustio-Mart{\'i}nez and Felipe Orihuela-Espina",

note = "Funding Information: The present work was supported by CONACyT/M{\'e}xico (schol-arship 937210 and grant CB-2015-01-257383) and Labex EFL through EFL mobility grants. Additionally, the authors thank CONACYT for the computer resources provided through the INAOE Supercomputing Laboratory{\textquoteright}s Deep Learning Platform for Language Technologies. Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 21st Mexican International Conference on Artificial Intelligence, MICAI 2022 ; Conference date: 24-10-2022 Through 29-10-2022",

year = "2022",

month = oct,

day = "25",

doi = "10.1007/978-3-031-19496-2_8",

language = "English",

isbn = "9783031194955",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "101--113",

editor = "{Pichardo Lagunas}, Obdulia and {Mart{\'i}nez Seis}, Bella and Juan Mart{\'i}nez-Miranda",

booktitle = "Advances in Computational Intelligence",

edition = "1",

}

García-Mendoza, JL, Villaseñor-Pineda, L, Buscaldi, D, Bustio-Martínez, L & Orihuela-Espina, F 2022, Evaluation of a new representation for noise reduction in distant supervision. in O Pichardo Lagunas, B Martínez Seis & J Martínez-Miranda (eds), Advances in Computational Intelligence: 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II. 1 edn, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13613 LNAI, Springer, pp. 101-113, 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Monterrey, Mexico, 24/10/22. https://doi.org/10.1007/978-3-031-19496-2_8

Evaluation of a new representation for noise reduction in distant supervision. / García-Mendoza, Juan Luis; Villaseñor-Pineda, Luis; Buscaldi, Davide et al.
Advances in Computational Intelligence: 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II. ed. / Obdulia Pichardo Lagunas; Bella Martínez Seis; Juan Martínez-Miranda. 1. ed. Springer, 2022. p. 101-113 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13613 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Evaluation of a new representation for noise reduction in distant supervision

AU - García-Mendoza, Juan Luis

AU - Villaseñor-Pineda, Luis

AU - Buscaldi, Davide

AU - Bustio-Martínez, Lázaro

AU - Orihuela-Espina, Felipe

N1 - Funding Information: The present work was supported by CONACyT/México (schol-arship 937210 and grant CB-2015-01-257383) and Labex EFL through EFL mobility grants. Additionally, the authors thank CONACYT for the computer resources provided through the INAOE Supercomputing Laboratory’s Deep Learning Platform for Language Technologies. Publisher Copyright: © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

PY - 2022/10/25

Y1 - 2022/10/25

N2 - Distant Supervision is a relation extraction approach that allows automatic labeling of a dataset. However, this labeling introduces noise in the labels (e.g., when two entities in a sentence are automatically labeled with an invalid relation). Noise in labels makes difficult the relation extraction task. This noise is precisely one of the main challenges of this task. Until now, the methods that incorporate a previous noise reduction step do not evaluate the performance of this step. This paper evaluates the noise reduction using a new representation obtained with autoencoders. In addition, it was incoporated more information to the input of the autoencoder proposed in the state-of-the-art to improve the representation over which the noise is reduced. Also, three methods were proposed to select the instances considered as real. As a result, it was obtained the highest values of the area under the ROC curves using the improved input combined with state-of-the-art anomaly detection methods. Moreover, the three proposed selection methods significantly improve the existing method in the literature.

AB - Distant Supervision is a relation extraction approach that allows automatic labeling of a dataset. However, this labeling introduces noise in the labels (e.g., when two entities in a sentence are automatically labeled with an invalid relation). Noise in labels makes difficult the relation extraction task. This noise is precisely one of the main challenges of this task. Until now, the methods that incorporate a previous noise reduction step do not evaluate the performance of this step. This paper evaluates the noise reduction using a new representation obtained with autoencoders. In addition, it was incoporated more information to the input of the autoencoder proposed in the state-of-the-art to improve the representation over which the noise is reduced. Also, three methods were proposed to select the instances considered as real. As a result, it was obtained the highest values of the area under the ROC curves using the improved input combined with state-of-the-art anomaly detection methods. Moreover, the three proposed selection methods significantly improve the existing method in the literature.

KW - Adversarial autoencoders

KW - Data representation

KW - Distant supervision

KW - Noise reduction

UR - http://www.scopus.com/inward/record.url?scp=85142848481&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-19496-2_8

DO - 10.1007/978-3-031-19496-2_8

M3 - Conference contribution

AN - SCOPUS:85142848481

SN - 9783031194955

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 101

EP - 113

BT - Advances in Computational Intelligence

A2 - Pichardo Lagunas, Obdulia

A2 - Martínez Seis, Bella

A2 - Martínez-Miranda, Juan

PB - Springer

T2 - 21st Mexican International Conference on Artificial Intelligence, MICAI 2022

Y2 - 24 October 2022 through 29 October 2022

ER -

García-Mendoza JL, Villaseñor-Pineda L, Buscaldi D, Bustio-Martínez L, Orihuela-Espina F. Evaluation of a new representation for noise reduction in distant supervision. In Pichardo Lagunas O, Martínez Seis B, Martínez-Miranda J, editors, Advances in Computational Intelligence: 21st Mexican International Conference on Artificial Intelligence, MICAI 2022, Proceedings, Part II. 1 ed. Springer. 2022. p. 101-113. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-19496-2_8

Evaluation of a new representation for noise reduction in distant supervision

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this