Comparação de desempenho do algoritmo Deep Q-Learning em ambientes simulados com estados contínuos
Reinforcement learning emerged in the 1980s and is one of three main areas of machine learning, the other two being supervised and unsupervised learning. Reinforcement problems have unique characteristics, such as the exchange of information between the agent and the environment in which it is inser...
| Autor principal: | Colombo, Gabriel |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2022
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/29123 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
Reinforcement learning emerged in the 1980s and is one of three main areas of machine learning, the other two being supervised and unsupervised learning. Reinforcement problems have unique characteristics, such as the exchange of information between the agent and the environment in which it is inserted. In addition, all reinforcement learning problems are based on objectives and make use of rewards as stimulus for learning. Another particularity of reinforcement learning is that it does not need prior information about the environment, as it is possible to collect data from interactions, using trial and error techniques. Although it emerged in the 1980s, reinforcement learning has recently gained popularity with the advancement of neural networks and the emergence of deep neural networks, since the fact that they can find function approximations has made it possible to solve problems with infinite states, which are more similar to problems in the real world. A major ambition of reinforcement learning is to create an algorithm that can be generalized and adapted to various environments. In this sense, this work aims to evaluate the Deep Q-Learning algorithm on 5 continuous state environments and to analyze both its performance and its adaptation capacity for different environments. |
|---|