Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies

Pieter Libin, Timothy Verstraeten, D.M. Roijers, Jelena Grujic, Kristof Theys, Phillippe Lemey, Ann Nowé

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

Abstract

Pandemic influenza has the epidemic potential to kill millions of people. While various preventive measures exist (i.a., vaccination and school closures), deciding on strategies that lead to their most effective and efficient use remains challenging. To this end, individual-based epidemiological models are essential to assist decision makers in determining the best strategy to curb epidemic spread. However, individual-based models are computationally intensive and it is therefore pivotal to identify the optimal strategy using a minimal amount of model evaluations. Additionally, as epidemiological modeling experiments need to be planned, a computational budget needs to be specified a priori. Consequently, we present a new sampling technique to optimize the evaluation of preventive strategies using fixed budget best-arm identification algorithms. We use epidemiological modeling theory to derive knowledge about the reward distribution which we exploit using Bayesian best-arm identification algorithms (i.e., Top-two Thompson sampling and BayesGap). We evaluate these algorithms in a realistic experimental setting and demonstrate that it is possible to identify the optimal strategy using only a limited number of model evaluations, i.e., 2-to-3 times faster compared to the uniform sampling method, the predominant technique used for epidemiological decision making in the literature. Finally, we contribute and evaluate a statistic for Top-two Thompson sampling to inform the decision makers about the confidence of an arm recommendation.
Original languageEnglish
Title of host publicationECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases
Subtitle of host publicationEuropean Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings
EditorsUlf Brefeld
PublisherSpringer
Pages456-471
Number of pages16
Volume3
ISBN (Electronic)9783030109974
ISBN (Print)9783030109967
DOIs
Publication statusPublished - 2019
EventECML PKDD 2018 - The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases - http://www.ecmlpkdd2018.org/, Dublin, Ireland
Duration: 10 Sep 2018 → …

Publication series

NameLecture Notes in Computer Science
PublisherSpringer
Volume11053

Conference

ConferenceECML PKDD 2018 - The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases
Abbreviated titleECML PKDD 2018
CountryIreland
CityDublin
Period10/09/18 → …

Fingerprint

Sampling
Curbs
Decision making
Statistics
Experiments

Cite this

Libin, P., Verstraeten, T., Roijers, D. M., Grujic, J., Theys, K., Lemey, P., & Nowé, A. (2019). Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies. In U. Brefeld (Ed.), ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings (Vol. 3, pp. 456-471 ). (Lecture Notes in Computer Science; Vol. 11053). Springer. https://doi.org/10.1007/978-3-030-10997-4_28
Libin, Pieter ; Verstraeten, Timothy ; Roijers, D.M. ; Grujic, Jelena ; Theys, Kristof ; Lemey, Phillippe ; Nowé, Ann. / Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies. ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings. editor / Ulf Brefeld. Vol. 3 Springer, 2019. pp. 456-471 (Lecture Notes in Computer Science).
@inproceedings{10951ade532b4ef4a83373de7d7663bb,
title = "Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies",
abstract = "Pandemic influenza has the epidemic potential to kill millions of people. While various preventive measures exist (i.a., vaccination and school closures), deciding on strategies that lead to their most effective and efficient use remains challenging. To this end, individual-based epidemiological models are essential to assist decision makers in determining the best strategy to curb epidemic spread. However, individual-based models are computationally intensive and it is therefore pivotal to identify the optimal strategy using a minimal amount of model evaluations. Additionally, as epidemiological modeling experiments need to be planned, a computational budget needs to be specified a priori. Consequently, we present a new sampling technique to optimize the evaluation of preventive strategies using fixed budget best-arm identification algorithms. We use epidemiological modeling theory to derive knowledge about the reward distribution which we exploit using Bayesian best-arm identification algorithms (i.e., Top-two Thompson sampling and BayesGap). We evaluate these algorithms in a realistic experimental setting and demonstrate that it is possible to identify the optimal strategy using only a limited number of model evaluations, i.e., 2-to-3 times faster compared to the uniform sampling method, the predominant technique used for epidemiological decision making in the literature. Finally, we contribute and evaluate a statistic for Top-two Thompson sampling to inform the decision makers about the confidence of an arm recommendation.",
author = "Pieter Libin and Timothy Verstraeten and D.M. Roijers and Jelena Grujic and Kristof Theys and Phillippe Lemey and Ann Now{\'e}",
year = "2019",
doi = "10.1007/978-3-030-10997-4_28",
language = "English",
isbn = "9783030109967",
volume = "3",
series = "Lecture Notes in Computer Science",
publisher = "Springer",
pages = "456--471",
editor = "Ulf Brefeld",
booktitle = "ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases",

}

Libin, P, Verstraeten, T, Roijers, DM, Grujic, J, Theys, K, Lemey, P & Nowé, A 2019, Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies. in U Brefeld (ed.), ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings. vol. 3, Lecture Notes in Computer Science, vol. 11053, Springer, pp. 456-471 , ECML PKDD 2018 - The European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, Dublin, Ireland, 10/09/18. https://doi.org/10.1007/978-3-030-10997-4_28

Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies. / Libin, Pieter; Verstraeten, Timothy; Roijers, D.M.; Grujic, Jelena; Theys, Kristof; Lemey, Phillippe; Nowé, Ann.

ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings. ed. / Ulf Brefeld. Vol. 3 Springer, 2019. p. 456-471 (Lecture Notes in Computer Science; Vol. 11053).

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

TY - GEN

T1 - Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies

AU - Libin, Pieter

AU - Verstraeten, Timothy

AU - Roijers, D.M.

AU - Grujic, Jelena

AU - Theys, Kristof

AU - Lemey, Phillippe

AU - Nowé, Ann

PY - 2019

Y1 - 2019

N2 - Pandemic influenza has the epidemic potential to kill millions of people. While various preventive measures exist (i.a., vaccination and school closures), deciding on strategies that lead to their most effective and efficient use remains challenging. To this end, individual-based epidemiological models are essential to assist decision makers in determining the best strategy to curb epidemic spread. However, individual-based models are computationally intensive and it is therefore pivotal to identify the optimal strategy using a minimal amount of model evaluations. Additionally, as epidemiological modeling experiments need to be planned, a computational budget needs to be specified a priori. Consequently, we present a new sampling technique to optimize the evaluation of preventive strategies using fixed budget best-arm identification algorithms. We use epidemiological modeling theory to derive knowledge about the reward distribution which we exploit using Bayesian best-arm identification algorithms (i.e., Top-two Thompson sampling and BayesGap). We evaluate these algorithms in a realistic experimental setting and demonstrate that it is possible to identify the optimal strategy using only a limited number of model evaluations, i.e., 2-to-3 times faster compared to the uniform sampling method, the predominant technique used for epidemiological decision making in the literature. Finally, we contribute and evaluate a statistic for Top-two Thompson sampling to inform the decision makers about the confidence of an arm recommendation.

AB - Pandemic influenza has the epidemic potential to kill millions of people. While various preventive measures exist (i.a., vaccination and school closures), deciding on strategies that lead to their most effective and efficient use remains challenging. To this end, individual-based epidemiological models are essential to assist decision makers in determining the best strategy to curb epidemic spread. However, individual-based models are computationally intensive and it is therefore pivotal to identify the optimal strategy using a minimal amount of model evaluations. Additionally, as epidemiological modeling experiments need to be planned, a computational budget needs to be specified a priori. Consequently, we present a new sampling technique to optimize the evaluation of preventive strategies using fixed budget best-arm identification algorithms. We use epidemiological modeling theory to derive knowledge about the reward distribution which we exploit using Bayesian best-arm identification algorithms (i.e., Top-two Thompson sampling and BayesGap). We evaluate these algorithms in a realistic experimental setting and demonstrate that it is possible to identify the optimal strategy using only a limited number of model evaluations, i.e., 2-to-3 times faster compared to the uniform sampling method, the predominant technique used for epidemiological decision making in the literature. Finally, we contribute and evaluate a statistic for Top-two Thompson sampling to inform the decision makers about the confidence of an arm recommendation.

UR - https://arxiv.org/pdf/1711.06299.pdf

U2 - 10.1007/978-3-030-10997-4_28

DO - 10.1007/978-3-030-10997-4_28

M3 - Conference contribution

SN - 9783030109967

VL - 3

T3 - Lecture Notes in Computer Science

SP - 456

EP - 471

BT - ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases

A2 - Brefeld, Ulf

PB - Springer

ER -

Libin P, Verstraeten T, Roijers DM, Grujic J, Theys K, Lemey P et al. Bayesian Best-Arm Identification for Selecting Influenza Mitigation Strategies. In Brefeld U, editor, ECML PKDD 2018 Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2018, Dublin, Ireland, September 10–14, 2018, Proceedings. Vol. 3. Springer. 2019. p. 456-471 . (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-030-10997-4_28