Narrowing reinforcement learning: Overcoming the cold start problem for personalized health interventions

Seyed Amin Tabatabaei*, Mark Hoogendoorn, Aart van Halteren

*Corresponding author for this work

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

Abstract

Personalization of support in health and wellbeing settings is challenging. While personalization has shown to be highly beneficial to maximize the success of interventions, often only very limited experiences are available to personalize support strategies. Because of its focus on finding suitable actions/interventions that lead to long term rewards, reinforcement learning is very suitable for personalization but requires a substantial learning period. To overcome this so-called cold start problem, we propose a novel approach called narrowing reinforcement learning. The approach exploits experiences of the nearest neighbors around a user to generate a suitable policy, expressing which action to perform in what state. Using a narrowing function, the size of the neighborhood is reduced as more experiences are collected, allowing for the most personalized experience that is possible given the amount of collected experiences. An evaluation of the approach in a realistic simulator shows that it significantly outperforms the current state-of-the-art approaches for personalization in health and wellbeing using reinforcement learning.

Original languageEnglish
Title of host publicationPRIMA 2018: Principles and Practice of Multi-Agent Systems
Subtitle of host publication21st International Conference, Tokyo, Japan, October 29-November 2, 2018, Proceedings
EditorsNir Oren, Yuko Sakurai, Itsuki Noda, Tran Cao Son, Tim Miller, Bastin Tony Savarimuthu
PublisherSpringer/Verlag
Chapter19
Pages312-327
Number of pages16
ISBN (Electronic)9783030030988
ISBN (Print)9783030030971
DOIs
Publication statusPublished - 2018
Event21st International Conference on Principles and Practice of Multi-Agent Systems, PRIMA 2018 - Tokyo, Japan
Duration: 29 Oct 20182 Nov 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11224 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference21st International Conference on Principles and Practice of Multi-Agent Systems, PRIMA 2018
CountryJapan
CityTokyo
Period29/10/182/11/18

Keywords

  • Health
  • Personalization
  • Reinforcement learning

Fingerprint Dive into the research topics of 'Narrowing reinforcement learning: Overcoming the cold start problem for personalized health interventions'. Together they form a unique fingerprint.

Cite this