Using generative adversarial networks to develop a realistic human behavior simulator

Ali el Hassouni*, Mark Hoogendoorn, Vesa Muhonen

*Corresponding author for this work

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

Abstract

Simulation environments have proven to be very useful as testbeds for reinforcement learning (RL) algorithms. For settings where an actual human user is involved, these simulation environments allow one to test out the suitability of new RL approaches without having to include real users at first. It obviously does require the simulator to have a certain degree of realism, however, realistic simulators for the behavior of humans in the health domain are rarely seen. To generate realistic behavior, the simulator could be driven by data from real users, but this might lead to privacy issues. In this paper, we propose to use Generative Adversarial Networks (GANs) for generating realistic simulation environments. In this first step, we use an existing simulator that simulates daily activities of users and the GANs are used to generate realistic sensory data that accompanies such activities. After training, the original (potentially privacy sensitive) data can be thrown away and the simulator can simply be driven by the GAN models. Results show that a model trained on real data shows similar performance on the data artificially generated by the GAN.

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
Pages476-483
Number of pages8
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

  • Simulation
  • Generative adversarial networks
  • Reinforcement learning
  • Deep learning
  • e-Health

Fingerprint Dive into the research topics of 'Using generative adversarial networks to develop a realistic human behavior simulator'. Together they form a unique fingerprint.

Cite this