Mueller-Hof, Niclas (2024) Improving Efficiency of a Hierarchical Reinforcement Learning Algorithm. Bachelor's Thesis, Artificial Intelligence.
|
Text
bAI2024Mueller-HofNJ.pdf Download (872kB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (155kB) |
Abstract
This thesis explores the efficiency of a hierarchical reinforcement learning algorithm using a fixed, state-dependent compression function to manage complex tasks with sparse rewards. To enhance learning efficiency, a multi-headed neural network architecture was proposed, enabling parameter sharing across subtasks while maintaining specialized outputs. However, experimental results indicated that this approach did not outperform the single neural network for each option, likely due to overgeneralization and insufficient capacity in shared layers. The study suggests future research should also focus on improving the multi-headed architecture to better balance shared and specialized components, potentially enhancing flexibility and overall performance.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Fernandes Cunha, R. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 13 Aug 2024 09:46 |
| Last Modified: | 13 Aug 2024 09:46 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/33952 |
Actions (login required)
 |
View Item |