Active Learning for Player Modeling

Noor Shaker, Mohamed Abou-Zleikha, Mohammad Shaker

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

Abstract

Learning models of player behavior has been the focus of several studies. This work is motivated by better understanding of player behavior, a knowledge that can ultimately be employed to provide player-adapted or personalized content. In this paper, we propose the use of active learning for player experience modeling. We use a dataset from hundreds of players playing Infinite Mario Bros. as a case study and we employ the random forest method to learn mod- els of player experience through the active learning approach. The results obtained suggest that only part of the dataset (up to half the size of the full dataset) is necessary for the construction of accu- rate models that are as accurate as those constructed from the full dataset. This indicates the potential of the method and its benefits in cases when obtaining the data is expensive or time, storage or effort consuming. The results also indicate that the method can be used online during the content generation process where the mod- els can improve and better content can be presented as the game is being played.
Original languageEnglish
Title of host publicationProceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015)
Number of pages8
PublisherAssociation for Computing Machinery
Publication date2015
ISBN (Print)978-0-9913982-4-9
Publication statusPublished - 2015
EventFoundations of Digital Games Conference 2015 - Pacific Grove, CA, United States
Duration: 22 Jun 201525 Jun 2015

Conference

ConferenceFoundations of Digital Games Conference 2015
CountryUnited States
CityPacific Grove, CA
Period22/06/201525/06/2015

Cite this

Shaker, N., Abou-Zleikha, M., & Shaker, M. (2015). Active Learning for Player Modeling. In Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015) Association for Computing Machinery.
Shaker, Noor ; Abou-Zleikha, Mohamed ; Shaker, Mohammad. / Active Learning for Player Modeling. Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015). Association for Computing Machinery, 2015.
@inproceedings{e7d3fd85044f441999d88a5323b43df9,
title = "Active Learning for Player Modeling",
abstract = "Learning models of player behavior has been the focus of several studies. This work is motivated by better understanding of player behavior, a knowledge that can ultimately be employed to provide player-adapted or personalized content. In this paper, we propose the use of active learning for player experience modeling. We use a dataset from hundreds of players playing Infinite Mario Bros. as a case study and we employ the random forest method to learn mod- els of player experience through the active learning approach. The results obtained suggest that only part of the dataset (up to half the size of the full dataset) is necessary for the construction of accu- rate models that are as accurate as those constructed from the full dataset. This indicates the potential of the method and its benefits in cases when obtaining the data is expensive or time, storage or effort consuming. The results also indicate that the method can be used online during the content generation process where the mod- els can improve and better content can be presented as the game is being played.",
author = "Noor Shaker and Mohamed Abou-Zleikha and Mohammad Shaker",
year = "2015",
language = "English",
isbn = "978-0-9913982-4-9",
booktitle = "Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015)",
publisher = "Association for Computing Machinery",
address = "United States",

}

Shaker, N, Abou-Zleikha, M & Shaker, M 2015, Active Learning for Player Modeling. in Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015). Association for Computing Machinery, Foundations of Digital Games Conference 2015, Pacific Grove, CA, United States, 22/06/2015.

Active Learning for Player Modeling. / Shaker, Noor; Abou-Zleikha, Mohamed; Shaker, Mohammad.

Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015). Association for Computing Machinery, 2015.

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

TY - GEN

T1 - Active Learning for Player Modeling

AU - Shaker, Noor

AU - Abou-Zleikha, Mohamed

AU - Shaker, Mohammad

PY - 2015

Y1 - 2015

N2 - Learning models of player behavior has been the focus of several studies. This work is motivated by better understanding of player behavior, a knowledge that can ultimately be employed to provide player-adapted or personalized content. In this paper, we propose the use of active learning for player experience modeling. We use a dataset from hundreds of players playing Infinite Mario Bros. as a case study and we employ the random forest method to learn mod- els of player experience through the active learning approach. The results obtained suggest that only part of the dataset (up to half the size of the full dataset) is necessary for the construction of accu- rate models that are as accurate as those constructed from the full dataset. This indicates the potential of the method and its benefits in cases when obtaining the data is expensive or time, storage or effort consuming. The results also indicate that the method can be used online during the content generation process where the mod- els can improve and better content can be presented as the game is being played.

AB - Learning models of player behavior has been the focus of several studies. This work is motivated by better understanding of player behavior, a knowledge that can ultimately be employed to provide player-adapted or personalized content. In this paper, we propose the use of active learning for player experience modeling. We use a dataset from hundreds of players playing Infinite Mario Bros. as a case study and we employ the random forest method to learn mod- els of player experience through the active learning approach. The results obtained suggest that only part of the dataset (up to half the size of the full dataset) is necessary for the construction of accu- rate models that are as accurate as those constructed from the full dataset. This indicates the potential of the method and its benefits in cases when obtaining the data is expensive or time, storage or effort consuming. The results also indicate that the method can be used online during the content generation process where the mod- els can improve and better content can be presented as the game is being played.

M3 - Article in proceeding

SN - 978-0-9913982-4-9

BT - Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015)

PB - Association for Computing Machinery

ER -

Shaker N, Abou-Zleikha M, Shaker M. Active Learning for Player Modeling. In Proceedings of the 10th International Conference on the Foundations of Digital Games (FDG 2015). Association for Computing Machinery. 2015