Details
| Original language | English |
|---|---|
| Title of host publication | IEEE Conference on Games 2019, CoG 2019 |
| Publisher | IEEE Computer Society |
| ISBN (electronic) | 9781728118840 |
| Publication status | Published - Aug 2019 |
| Externally published | Yes |
| Event | 2019 IEEE Conference on Games, CoG 2019 - London, United Kingdom (UK) Duration: 20 Aug 2019 → 23 Aug 2019 |
Publication series
| Name | IEEE Conference on Computatonal Intelligence and Games, CIG |
|---|---|
| Volume | 2019-August |
| ISSN (Print) | 2325-4270 |
| ISSN (electronic) | 2325-4289 |
Abstract
This paper examines learning approaches for forward models based on local cell transition functions. We provide a formal definition of local forward models for which we propose two basic learning approaches. Our analysis is based on the game Sokoban, where a wrong action can lead to an unsolvable game state. Therefore, an accurate prediction of an action's resulting state is necessary to avoid this scenario.In contrast to learning the complete state transition function, local forward models allow extracting multiple training examples from a single state transition. In this way, the Hash Set model, as well as the Decision Tree model, quickly learn to predict upcoming state transitions of both the training and the test set. Applying the model using a statistical forward planner showed that the best models can be used to satisfying degree even in cases in which the test levels have not yet been seen.Our evaluation includes an analysis of various local neighbourhood patterns and sizes to test the learners' capabilities in case too few or too many attributes are extracted, of which the latter has shown do degrade the performance of the model learner.
Keywords
- Decision Tree, Forward Model Learning, Local Forward Model, Rolling Horizon Evolutionary Algorithm
ASJC Scopus subject areas
- Computer Science(all)
- Artificial Intelligence
- Computer Science(all)
- Computer Graphics and Computer-Aided Design
- Computer Science(all)
- Computer Vision and Pattern Recognition
- Computer Science(all)
- Human-Computer Interaction
- Computer Science(all)
- Software
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
IEEE Conference on Games 2019, CoG 2019. IEEE Computer Society, 2019. 8848044 (IEEE Conference on Computatonal Intelligence and Games, CIG; Vol. 2019-August).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Learning local forward models on unforgiving games
AU - Dockhorn, Alexander
AU - Lucas, Simon M.
AU - Volz, Vanessa
AU - Bravi, Ivan
AU - Gaina, Raluca D.
AU - Perez-Liebana, Diego
PY - 2019/8
Y1 - 2019/8
N2 - This paper examines learning approaches for forward models based on local cell transition functions. We provide a formal definition of local forward models for which we propose two basic learning approaches. Our analysis is based on the game Sokoban, where a wrong action can lead to an unsolvable game state. Therefore, an accurate prediction of an action's resulting state is necessary to avoid this scenario.In contrast to learning the complete state transition function, local forward models allow extracting multiple training examples from a single state transition. In this way, the Hash Set model, as well as the Decision Tree model, quickly learn to predict upcoming state transitions of both the training and the test set. Applying the model using a statistical forward planner showed that the best models can be used to satisfying degree even in cases in which the test levels have not yet been seen.Our evaluation includes an analysis of various local neighbourhood patterns and sizes to test the learners' capabilities in case too few or too many attributes are extracted, of which the latter has shown do degrade the performance of the model learner.
AB - This paper examines learning approaches for forward models based on local cell transition functions. We provide a formal definition of local forward models for which we propose two basic learning approaches. Our analysis is based on the game Sokoban, where a wrong action can lead to an unsolvable game state. Therefore, an accurate prediction of an action's resulting state is necessary to avoid this scenario.In contrast to learning the complete state transition function, local forward models allow extracting multiple training examples from a single state transition. In this way, the Hash Set model, as well as the Decision Tree model, quickly learn to predict upcoming state transitions of both the training and the test set. Applying the model using a statistical forward planner showed that the best models can be used to satisfying degree even in cases in which the test levels have not yet been seen.Our evaluation includes an analysis of various local neighbourhood patterns and sizes to test the learners' capabilities in case too few or too many attributes are extracted, of which the latter has shown do degrade the performance of the model learner.
KW - Decision Tree
KW - Forward Model Learning
KW - Local Forward Model
KW - Rolling Horizon Evolutionary Algorithm
UR - http://www.scopus.com/inward/record.url?scp=85073106140&partnerID=8YFLogxK
U2 - 10.1109/CIG.2019.8848044
DO - 10.1109/CIG.2019.8848044
M3 - Conference contribution
AN - SCOPUS:85073106140
T3 - IEEE Conference on Computatonal Intelligence and Games, CIG
BT - IEEE Conference on Games 2019, CoG 2019
PB - IEEE Computer Society
T2 - 2019 IEEE Conference on Games, CoG 2019
Y2 - 20 August 2019 through 23 August 2019
ER -