An Ensemble Approach to Time Dependent Classification

Florian van Daalen, Evgueni Smirnov, Nasser Davarzani, Ralf Peeters, Joël Karel, Hanspeter Brunner-La Rocca

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingAcademicpeer-review

Abstract

A difficult aspect of a time dependent classification task is that the data are not IID sampled. To model this dependency several approaches in longitudinal analysis were developed. However, these approaches either have trouble estimating their generalization performance or are parametric in a statistical sense. To overcome these problems we propose in this paper a new approach of time dependent ensembles. Our approach decomposes the time dependent classification task into a series of classification tasks with IID sampled data. Each task can be solved by a classifier that is not supposed to model any dependency in the data. This allows for the use of a much broader spectrum of existing approaches than is possible on the original data. The classifiers associated with the tasks form the time dependent ensembles. The ensembles estimates the final class of the objects being classified by using a voting scheme. The experiments show the potential of the time dependent ensembles.
Original languageEnglish
Title of host publication17th IEEE International Conference on Machine Learning and Applications
PublisherIEEE
Number of pages5
ISBN (Electronic)978-1-5386-6805-4
DOIs
Publication statusPublished - 20 Dec 2018

Cite this

van Daalen, F., Smirnov, E., Davarzani, N., Peeters, R., Karel, J., & Brunner-La Rocca, H. (2018). An Ensemble Approach to Time Dependent Classification. In 17th IEEE International Conference on Machine Learning and Applications IEEE. https://doi.org/10.1109/ICMLA.2018.00164
van Daalen, Florian ; Smirnov, Evgueni ; Davarzani, Nasser ; Peeters, Ralf ; Karel, Joël ; Brunner-La Rocca, Hanspeter. / An Ensemble Approach to Time Dependent Classification. 17th IEEE International Conference on Machine Learning and Applications. IEEE, 2018.
@inproceedings{5d0b9bc9a0c9422db138469a90e91a0c,
title = "An Ensemble Approach to Time Dependent Classification",
abstract = "A difficult aspect of a time dependent classification task is that the data are not IID sampled. To model this dependency several approaches in longitudinal analysis were developed. However, these approaches either have trouble estimating their generalization performance or are parametric in a statistical sense. To overcome these problems we propose in this paper a new approach of time dependent ensembles. Our approach decomposes the time dependent classification task into a series of classification tasks with IID sampled data. Each task can be solved by a classifier that is not supposed to model any dependency in the data. This allows for the use of a much broader spectrum of existing approaches than is possible on the original data. The classifiers associated with the tasks form the time dependent ensembles. The ensembles estimates the final class of the objects being classified by using a voting scheme. The experiments show the potential of the time dependent ensembles.",
author = "{van Daalen}, Florian and Evgueni Smirnov and Nasser Davarzani and Ralf Peeters and Jo{\"e}l Karel and {Brunner-La Rocca}, Hanspeter",
year = "2018",
month = "12",
day = "20",
doi = "10.1109/ICMLA.2018.00164",
language = "English",
booktitle = "17th IEEE International Conference on Machine Learning and Applications",
publisher = "IEEE",
address = "United States",

}

van Daalen, F, Smirnov, E, Davarzani, N, Peeters, R, Karel, J & Brunner-La Rocca, H 2018, An Ensemble Approach to Time Dependent Classification. in 17th IEEE International Conference on Machine Learning and Applications. IEEE. https://doi.org/10.1109/ICMLA.2018.00164

An Ensemble Approach to Time Dependent Classification. / van Daalen, Florian; Smirnov, Evgueni; Davarzani, Nasser; Peeters, Ralf; Karel, Joël; Brunner-La Rocca, Hanspeter.

17th IEEE International Conference on Machine Learning and Applications. IEEE, 2018.

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingAcademicpeer-review

TY - GEN

T1 - An Ensemble Approach to Time Dependent Classification

AU - van Daalen, Florian

AU - Smirnov, Evgueni

AU - Davarzani, Nasser

AU - Peeters, Ralf

AU - Karel, Joël

AU - Brunner-La Rocca, Hanspeter

PY - 2018/12/20

Y1 - 2018/12/20

N2 - A difficult aspect of a time dependent classification task is that the data are not IID sampled. To model this dependency several approaches in longitudinal analysis were developed. However, these approaches either have trouble estimating their generalization performance or are parametric in a statistical sense. To overcome these problems we propose in this paper a new approach of time dependent ensembles. Our approach decomposes the time dependent classification task into a series of classification tasks with IID sampled data. Each task can be solved by a classifier that is not supposed to model any dependency in the data. This allows for the use of a much broader spectrum of existing approaches than is possible on the original data. The classifiers associated with the tasks form the time dependent ensembles. The ensembles estimates the final class of the objects being classified by using a voting scheme. The experiments show the potential of the time dependent ensembles.

AB - A difficult aspect of a time dependent classification task is that the data are not IID sampled. To model this dependency several approaches in longitudinal analysis were developed. However, these approaches either have trouble estimating their generalization performance or are parametric in a statistical sense. To overcome these problems we propose in this paper a new approach of time dependent ensembles. Our approach decomposes the time dependent classification task into a series of classification tasks with IID sampled data. Each task can be solved by a classifier that is not supposed to model any dependency in the data. This allows for the use of a much broader spectrum of existing approaches than is possible on the original data. The classifiers associated with the tasks form the time dependent ensembles. The ensembles estimates the final class of the objects being classified by using a voting scheme. The experiments show the potential of the time dependent ensembles.

U2 - 10.1109/ICMLA.2018.00164

DO - 10.1109/ICMLA.2018.00164

M3 - Conference article in proceeding

BT - 17th IEEE International Conference on Machine Learning and Applications

PB - IEEE

ER -

van Daalen F, Smirnov E, Davarzani N, Peeters R, Karel J, Brunner-La Rocca H. An Ensemble Approach to Time Dependent Classification. In 17th IEEE International Conference on Machine Learning and Applications. IEEE. 2018 https://doi.org/10.1109/ICMLA.2018.00164