Dimofte, Alexandru (2022) On the existence of Early-Bird tickets in Reinforcement Learning and how to find them. Bachelor's Thesis, Artificial Intelligence.
|
Text
Thesis_ADimofte.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 (112kB) |
Abstract
The Winning Lottery ticket hypothesis has shown that a neural network can contain sparse sub-networks that perform similarly or better in comparison. This finding was extended by the Early-Bird ticket hypothesis, which revealed that winning lottery tickets can be found early within the first few initial training steps. This study demonstrates the existence of EarlyBird tickets in neural networks used in reinforcement learning and introduces EB Exploit - an algorithm which can reliably find EB tickets in simple RL problems, as well as increasing efficiency in the training process of such agents.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Sabatelli, M. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 27 Jul 2022 07:17 |
| Last Modified: | 03 Feb 2023 09:34 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/27813 |
Actions (login required)
 |
View Item |