Stolt Ansó, Nil (2018) Investigating state representations in deep reinforcement learning for pellet eating in Agar.io. Bachelor's Thesis, Artificial Intelligence.
|
Text
AI_BA_2018_NilStoltAnso.pdf Download (537kB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (94kB) |
Abstract
The online game Agar.io has become massively popular on the internet due to its intuitive game design and its ability to instantly match players with others around the world. The game has a continuous input and action space and allows to have diverse agents with complex strategies compete against each other. This paper first investigates how different state representations influence the learning process of a Q-learning algorithm. The representations examined range from raw pixel values to extracted handcrafted feature vision grids. Secondly, we investigate how different value function network architectures compare in performance. The architectures examined are two convolutional Deep Q-networks (DQN) of varying depth and one smaller multilayer perceptron (MLP). The results show that the Q-learning algorithm, together with prioritized experience replay, is able to play quite well. Handcrafted feature vision grids seem to require minimal resolution and network complexity, and outperform raw pixel input for all conditions and tasks tested.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Wiering, M.A. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 31 Jul 2018 |
| Last Modified: | 01 Aug 2018 12:43 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/18176 |
Actions (login required)
 |
View Item |