Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption

Joel García; Erik Zamora; Humberto Sossa

doi:10.1007/978-3-030-04491-6_8

Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption

Joel García, Erik Zamora, Humberto Sossa

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

4 Scopus citations

Abstract

Households are responsible for more than 40% of the global electricity consumption [7]. The analysis of this consumption to find unexpected behaviours could have a great impact on saving electricity. This research presents an experimental study of supervised and unsupervised neural networks for anomaly detection in electrical consumption. Multilayer perceptrons and autoencoders are used for each approach, respectively. In order to select the most suitable neural model in each case, there is a comparison of various architectures. The proposed methods are evaluated using real-world data from an individual home electric power usage dataset. The performance is compared with a traditional statistical procedure. Experiments show that the supervised approach has a significant improvement in anomaly detection rate. We evaluate different possible feature sets. The results demonstrate that temporal data and measures of consumption patterns such as mean, standard deviation and percentiles are necessary to achieve higher accuracy.

Original language	English
Title of host publication	Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings
Editors	María de Lourdes Martínez-Villaseñor, Ildar Batyrshin, Hiram Eredín Ponce Espinosa
Publisher	Springer Verlag
Pages	98-109
Number of pages	12
ISBN (Print)	9783030044909
DOIs	https://doi.org/10.1007/978-3-030-04491-6_8
State	Published - 2018
Event	17th Mexican International Conference on Artificial Intelligence, MICAI 2018 - Guadalajara, Mexico Duration: 22 Oct 2018 → 27 Oct 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	11288 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	17th Mexican International Conference on Artificial Intelligence, MICAI 2018
Country/Territory	Mexico
City	Guadalajara
Period	22/10/18 → 27/10/18

Keywords

Anomaly detection
Neural networks
Statistic
Supervised
Unsupervised

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-030-04491-6_8

Cite this

García, J., Zamora, E., & Sossa, H. (2018). Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption. In M. de Lourdes Martínez-Villaseñor, I. Batyrshin, & H. E. Ponce Espinosa (Eds.), Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings (pp. 98-109). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11288 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-030-04491-6_8

García, Joel ; Zamora, Erik ; Sossa, Humberto. / Supervised and unsupervised neural networks : Experimental study for anomaly detection in electrical consumption. Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings. editor / María de Lourdes Martínez-Villaseñor ; Ildar Batyrshin ; Hiram Eredín Ponce Espinosa. Springer Verlag, 2018. pp. 98-109 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{97fa4c3b4f454bb38b046b2602fbb840,

title = "Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption",

abstract = "Households are responsible for more than 40% of the global electricity consumption [7]. The analysis of this consumption to find unexpected behaviours could have a great impact on saving electricity. This research presents an experimental study of supervised and unsupervised neural networks for anomaly detection in electrical consumption. Multilayer perceptrons and autoencoders are used for each approach, respectively. In order to select the most suitable neural model in each case, there is a comparison of various architectures. The proposed methods are evaluated using real-world data from an individual home electric power usage dataset. The performance is compared with a traditional statistical procedure. Experiments show that the supervised approach has a significant improvement in anomaly detection rate. We evaluate different possible feature sets. The results demonstrate that temporal data and measures of consumption patterns such as mean, standard deviation and percentiles are necessary to achieve higher accuracy.",

keywords = "Anomaly detection, Neural networks, Statistic, Supervised, Unsupervised",

author = "Joel Garc{\'i}a and Erik Zamora and Humberto Sossa",

note = "Publisher Copyright: {\textcopyright} 2018, Springer Nature Switzerland AG.; 17th Mexican International Conference on Artificial Intelligence, MICAI 2018 ; Conference date: 22-10-2018 Through 27-10-2018",

year = "2018",

doi = "10.1007/978-3-030-04491-6_8",

language = "Ingl{\'e}s",

isbn = "9783030044909",

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

publisher = "Springer Verlag",

pages = "98--109",

editor = "{de Lourdes Mart{\'i}nez-Villase{\~n}or}, Mar{\'i}a and Ildar Batyrshin and {Ponce Espinosa}, {Hiram Ered{\'i}n}",

booktitle = "Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings",

address = "Alemania",

}

García, J, Zamora, E & Sossa, H 2018, Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption. in M de Lourdes Martínez-Villaseñor, I Batyrshin & HE Ponce Espinosa (eds), Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11288 LNAI, Springer Verlag, pp. 98-109, 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Guadalajara, Mexico, 22/10/18. https://doi.org/10.1007/978-3-030-04491-6_8

Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption. / García, Joel; Zamora, Erik ; Sossa, Humberto.
Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings. ed. / María de Lourdes Martínez-Villaseñor; Ildar Batyrshin; Hiram Eredín Ponce Espinosa. Springer Verlag, 2018. p. 98-109 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11288 LNAI).

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

TY - GEN

T1 - Supervised and unsupervised neural networks

T2 - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018

AU - García, Joel

AU - Zamora, Erik

AU - Sossa, Humberto

PY - 2018

Y1 - 2018

N2 - Households are responsible for more than 40% of the global electricity consumption [7]. The analysis of this consumption to find unexpected behaviours could have a great impact on saving electricity. This research presents an experimental study of supervised and unsupervised neural networks for anomaly detection in electrical consumption. Multilayer perceptrons and autoencoders are used for each approach, respectively. In order to select the most suitable neural model in each case, there is a comparison of various architectures. The proposed methods are evaluated using real-world data from an individual home electric power usage dataset. The performance is compared with a traditional statistical procedure. Experiments show that the supervised approach has a significant improvement in anomaly detection rate. We evaluate different possible feature sets. The results demonstrate that temporal data and measures of consumption patterns such as mean, standard deviation and percentiles are necessary to achieve higher accuracy.

AB - Households are responsible for more than 40% of the global electricity consumption [7]. The analysis of this consumption to find unexpected behaviours could have a great impact on saving electricity. This research presents an experimental study of supervised and unsupervised neural networks for anomaly detection in electrical consumption. Multilayer perceptrons and autoencoders are used for each approach, respectively. In order to select the most suitable neural model in each case, there is a comparison of various architectures. The proposed methods are evaluated using real-world data from an individual home electric power usage dataset. The performance is compared with a traditional statistical procedure. Experiments show that the supervised approach has a significant improvement in anomaly detection rate. We evaluate different possible feature sets. The results demonstrate that temporal data and measures of consumption patterns such as mean, standard deviation and percentiles are necessary to achieve higher accuracy.

KW - Anomaly detection

KW - Neural networks

KW - Statistic

KW - Supervised

KW - Unsupervised

UR - http://www.scopus.com/inward/record.url?scp=85059934282&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-04491-6_8

DO - 10.1007/978-3-030-04491-6_8

M3 - Contribución a la conferencia

SN - 9783030044909

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

SP - 98

EP - 109

BT - Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings

A2 - de Lourdes Martínez-Villaseñor, María

A2 - Batyrshin, Ildar

A2 - Ponce Espinosa, Hiram Eredín

PB - Springer Verlag

Y2 - 22 October 2018 through 27 October 2018

ER -

García J, Zamora E , Sossa H. Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption. In de Lourdes Martínez-Villaseñor M, Batyrshin I, Ponce Espinosa HE, editors, Advances in Soft Computing - 17th Mexican International Conference on Artificial Intelligence, MICAI 2018, Proceedings. Springer Verlag. 2018. p. 98-109. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-04491-6_8

Supervised and unsupervised neural networks: Experimental study for anomaly detection in electrical consumption

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this