Modeling concept drift: A probabilistic graphical model based approach

Hanen Borchani, Ana Maria Martinez, Andrés R. Masegosa, Helge Langseth, Thomas Dyhre Nielsen, Antonio Salmerón, Antonio Fernández, Anders Læsø Madsen, Ramón Sáez

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

7 Citations (Scopus)
390 Downloads (Pure)

Abstract

An often used approach for detecting and adapting to concept drift when doing classification is to treat the data as i.i.d. and use changes in classification accuracy as an indication of concept drift. In this paper, we take a different perspective and propose a framework, based on probabilistic graphical models, that explicitly represents concept drift using latent variables. To ensure efficient inference and learning, we re- sort to a variational Bayes inference scheme. As a proof of concept, we demonstrate and analyze the proposed framework using synthetic data sets as well as a real financial data set from a Spanish bank.
Original languageEnglish
Title of host publicationAdvances in Intelligent Data Analysis XIV : 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings
EditorsElisa Fromont, Tijl De Bie, Matthijs van Leeuwen
PublisherSpringer
Publication date2015
Pages72-83
ISBN (Print)978-3-319-24464-8
ISBN (Electronic)978-3-319-24465-5
DOIs
Publication statusPublished - 2015
EventInternational Symposium, IDA 2015 - Saint Etienne, France
Duration: 22 Oct 201524 Oct 2015
Conference number: 14th

Conference

ConferenceInternational Symposium, IDA 2015
Number14th
CountryFrance
CitySaint Etienne
Period22/10/201524/10/2015
SeriesLecture Notes in Computer Science
Number9385
ISSN0302-9743

Cite this

Borchani, H., Martinez, A. M., Masegosa, A. R., Langseth, H., Nielsen, T. D., Salmerón, A., ... Sáez, R. (2015). Modeling concept drift: A probabilistic graphical model based approach. In E. Fromont, T. De Bie, & M. van Leeuwen (Eds.), Advances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings (pp. 72-83). Springer. Lecture Notes in Computer Science, No. 9385 https://doi.org/10.1007/978-3-319-24465-5_7
Borchani, Hanen ; Martinez, Ana Maria ; Masegosa, Andrés R. ; Langseth, Helge ; Nielsen, Thomas Dyhre ; Salmerón, Antonio ; Fernández, Antonio ; Madsen, Anders Læsø ; Sáez, Ramón. / Modeling concept drift : A probabilistic graphical model based approach. Advances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings. editor / Elisa Fromont ; Tijl De Bie ; Matthijs van Leeuwen. Springer, 2015. pp. 72-83 (Lecture Notes in Computer Science; No. 9385).
@inproceedings{e79a299c4f5a4df98b246a035472af30,
title = "Modeling concept drift: A probabilistic graphical model based approach",
abstract = "An often used approach for detecting and adapting to concept drift when doing classification is to treat the data as i.i.d. and use changes in classification accuracy as an indication of concept drift. In this paper, we take a different perspective and propose a framework, based on probabilistic graphical models, that explicitly represents concept drift using latent variables. To ensure efficient inference and learning, we re- sort to a variational Bayes inference scheme. As a proof of concept, we demonstrate and analyze the proposed framework using synthetic data sets as well as a real financial data set from a Spanish bank.",
author = "Hanen Borchani and Martinez, {Ana Maria} and Masegosa, {Andr{\'e}s R.} and Helge Langseth and Nielsen, {Thomas Dyhre} and Antonio Salmer{\'o}n and Antonio Fern{\'a}ndez and Madsen, {Anders L{\ae}s{\o}} and Ram{\'o}n S{\'a}ez",
year = "2015",
doi = "10.1007/978-3-319-24465-5_7",
language = "English",
isbn = "978-3-319-24464-8",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
number = "9385",
pages = "72--83",
editor = "Elisa Fromont and {De Bie}, Tijl and {van Leeuwen}, Matthijs",
booktitle = "Advances in Intelligent Data Analysis XIV",
address = "Germany",

}

Borchani, H, Martinez, AM, Masegosa, AR, Langseth, H, Nielsen, TD, Salmerón, A, Fernández, A, Madsen, AL & Sáez, R 2015, Modeling concept drift: A probabilistic graphical model based approach. in E Fromont, T De Bie & M van Leeuwen (eds), Advances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings. Springer, Lecture Notes in Computer Science, no. 9385, pp. 72-83, International Symposium, IDA 2015, Saint Etienne, France, 22/10/2015. https://doi.org/10.1007/978-3-319-24465-5_7

Modeling concept drift : A probabilistic graphical model based approach. / Borchani, Hanen; Martinez, Ana Maria; Masegosa, Andrés R.; Langseth, Helge; Nielsen, Thomas Dyhre; Salmerón, Antonio; Fernández, Antonio; Madsen, Anders Læsø; Sáez, Ramón.

Advances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings. ed. / Elisa Fromont; Tijl De Bie; Matthijs van Leeuwen. Springer, 2015. p. 72-83 (Lecture Notes in Computer Science; No. 9385).

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - Modeling concept drift

T2 - A probabilistic graphical model based approach

AU - Borchani, Hanen

AU - Martinez, Ana Maria

AU - Masegosa, Andrés R.

AU - Langseth, Helge

AU - Nielsen, Thomas Dyhre

AU - Salmerón, Antonio

AU - Fernández, Antonio

AU - Madsen, Anders Læsø

AU - Sáez, Ramón

PY - 2015

Y1 - 2015

N2 - An often used approach for detecting and adapting to concept drift when doing classification is to treat the data as i.i.d. and use changes in classification accuracy as an indication of concept drift. In this paper, we take a different perspective and propose a framework, based on probabilistic graphical models, that explicitly represents concept drift using latent variables. To ensure efficient inference and learning, we re- sort to a variational Bayes inference scheme. As a proof of concept, we demonstrate and analyze the proposed framework using synthetic data sets as well as a real financial data set from a Spanish bank.

AB - An often used approach for detecting and adapting to concept drift when doing classification is to treat the data as i.i.d. and use changes in classification accuracy as an indication of concept drift. In this paper, we take a different perspective and propose a framework, based on probabilistic graphical models, that explicitly represents concept drift using latent variables. To ensure efficient inference and learning, we re- sort to a variational Bayes inference scheme. As a proof of concept, we demonstrate and analyze the proposed framework using synthetic data sets as well as a real financial data set from a Spanish bank.

U2 - 10.1007/978-3-319-24465-5_7

DO - 10.1007/978-3-319-24465-5_7

M3 - Article in proceeding

SN - 978-3-319-24464-8

T3 - Lecture Notes in Computer Science

SP - 72

EP - 83

BT - Advances in Intelligent Data Analysis XIV

A2 - Fromont, Elisa

A2 - De Bie, Tijl

A2 - van Leeuwen, Matthijs

PB - Springer

ER -

Borchani H, Martinez AM, Masegosa AR, Langseth H, Nielsen TD, Salmerón A et al. Modeling concept drift: A probabilistic graphical model based approach. In Fromont E, De Bie T, van Leeuwen M, editors, Advances in Intelligent Data Analysis XIV: 14th International Symposium, IDA 2015, Saint Etienne. France, October 22 -24, 2015. Proceedings. Springer. 2015. p. 72-83. (Lecture Notes in Computer Science; No. 9385). https://doi.org/10.1007/978-3-319-24465-5_7