NEAT for video game learning: advancing agent intelligence through evolutionary algorithms
This document presents an exploration of NEAT (NeuroEvolution of Augmenting Topologies) as a powerful approach for training video game agents, with a focus on its application and effectiveness in a specific game. NEAT is a neuroevolutionary algorithm that combines artificial neural networks and gene...
- Autores:
-
Castellamos Matamoros, Boris
- Tipo de recurso:
- Trabajo de grado de pregrado
- Fecha de publicación:
- 2023
- Institución:
- Universidad de los Andes
- Repositorio:
- Séneca: repositorio Uniandes
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.uniandes.edu.co:1992/73636
- Acceso en línea:
- https://hdl.handle.net/1992/73636
- Palabra clave:
- Neat
Galaxian
Video game
Artificial intelligence
Ingeniería
- Rights
- openAccess
- License
- Attribution-ShareAlike 4.0 International
| Summary: | This document presents an exploration of NEAT (NeuroEvolution of Augmenting Topologies) as a powerful approach for training video game agents, with a focus on its application and effectiveness in a specific game. NEAT is a neuroevolutionary algorithm that combines artificial neural networks and genetic algorithms to evolve efficient neural networks capable of solving complex tasks. By dynamically adjusting network structures and connections, NEAT enables the discovery of novel gameplay strategies. Through a series of experiments and analysis in the context of the chosen game, this study aims to demonstrate the effectiveness of NEAT in optimizing agent behavior and achieving high levels of performance. Additionally, this document discusses potential future improvements and explores other potential applications of NEAT beyond video game AI, highlighting its versatility and potential for advancements in related fields. |
|---|