Neural Network based Gaming Project
Abstract Category: I.T.
Course / Degree: Software Engineering
Institution / University: Nottingham, United Kingdom
Published in: 2002
This project investigates and demonstrates the usefulness of neural network ‘reinforcement learning’ in autonomous computing objects within a gaming environment. TD-Learning will be the training schemes and the framework implemented for this simulation. A comparison between two Temporal Difference algorithms Q-learning and SARSA will be made, to see which fits best for the simulation. For each of the algorithms, different rates of the step-size parameter (alpha or α) will be tested to see which learning rate delivers the optimal learning performance.
The autonomous simulations have separate agents that work with different set objectives (e.g. avoiding maze walls and hazards, collecting objects, reaching the target). The agents will utilise reinforcement learning to achieve their goals.
Thesis Keywords/Search Tags:
Neural Network, Gaming
This Thesis Abstract may be cited as follows:
PATEL, A., 2002. Neural Network based Gaming Project, England, 2002
Submission Details: Thesis Abstract submitted by Asif Patel from United Kingdom on 18-Aug-2004 00:33.
Abstract has been viewed 3101 times (since 7 Mar 2010).
Asif Patel Contact Details: Email: asifpatel@spymac.com
Disclaimer
Great care has been taken to ensure that this information is correct, however ThesisAbstracts.com cannot accept responsibility for the contents of this Thesis abstract titled "Neural Network based Gaming Project". This abstract has been submitted by Asif Patel on 18-Aug-2004 00:33. You may report a problem using the contact form.
© Copyright 2003 - 2024 of ThesisAbstracts.com and respective owners.