Muscoi, Vlad-Nicolae (2024) Testing the performance of Asynchronous Advantage Actor Critic within environments featuring 2D and 3D graphics. Bachelor's Thesis, Artificial Intelligence.
|
Text
bAI2024MuscoiVN.pdf Download (1MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (160kB) |
Abstract
In this paper, we will use the popular games Super Mario Bros and Minecraft to test the performance of Asynchronous Advantage Actor-Critic (A3C), an algorithm that uses parallel agents to gather additional samples at the same time, within 2D and 3D environments respectively. Although high variance has been observed during training, the model when run within the Mario environment was able to converge to a policy capable of completing its task of finishing the level. Minecraft provided insight into the sample inefficiency of the algorithm, showing much slower learning for a much wider state space. The algorithm was not able to converge to a policy that would solve the task within the Minecraft environment. The code used within this project can be found on github: https://github.com/vluvl/A3C\_with\_Mario\_and\_Minecraft
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Cardenas Cartagena, J. D. and Sabatelli, M. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 01 Aug 2024 10:27 |
| Last Modified: | 01 Aug 2024 10:27 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/33795 |
Actions (login required)
 |
View Item |