Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Evolutionary algorithms for optimising reinforcement learning policy approximation
Reinforcement learning methods have become more efficient in recent years. In particular, the A3C (asynchronous advantage actor critic) approach demonstrated in Mnih et al. (2016) was able to halve the training time of the existing state-of-the-art approaches. However, these methods still require re...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Department of Statistical Sciences
2020
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613315831693312 |
|---|---|
| access_status_str | Open Access |
| author | Cuningham, Blake |
| author2 | Bassett, Bruce |
| author_browse | Bassett, Bruce Cuningham, Blake |
| author_facet | Bassett, Bruce Cuningham, Blake |
| author_sort | Cuningham, Blake |
| collection | Thesis |
| description | Reinforcement learning methods have become more efficient in recent years. In particular, the A3C (asynchronous advantage actor critic) approach demonstrated in Mnih et al. (2016) was able to halve the training time of the existing state-of-the-art approaches. However, these methods still require relatively large amounts of training resources due to the fundamental exploratory nature of reinforcement learning. Other machine learning approaches are able to improve the ability to train reinforcement learning agents by better processing input information to help map states to actions - convolutional and recurrent neural networks are helpful when input data is in image form that does not satisfy the Markov property. The specific required architecture of these convolutional and recurrent neural network models is not obvious given infinite possible permutations. There is very limited research giving clear guidance on neural network structure in a RL (reinforcement learning) context, and grid search-like approaches require too many resources and do not always find good optima. In order to address these, and other, challenges associated with traditional parameter optimization methods, an evolutionary approach similar to that taken by Dufourq and Bassett (2017) for image classification tasks was used to find the optimal model architecture when training an agent that learns to play Atari Pong. The approach found models that were able to train reinforcement learning agents faster, and with fewer parameters than that found by OpenAI’s model in Blackwell et al. (2018) - a superhuman level of performance. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/31170 |
| institution | University of Cape Town (South Africa) |
| language | eng |
| last_indexed | 2026-06-10T12:34:10.861Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2020 |
| publishDateRange | 2020 |
| publishDateSort | 2020 |
| publisher | Department of Statistical Sciences |
| publisherStr | Department of Statistical Sciences |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/31170 Evolutionary algorithms for optimising reinforcement learning policy approximation Cuningham, Blake Bassett, Bruce statistical sciences Reinforcement learning methods have become more efficient in recent years. In particular, the A3C (asynchronous advantage actor critic) approach demonstrated in Mnih et al. (2016) was able to halve the training time of the existing state-of-the-art approaches. However, these methods still require relatively large amounts of training resources due to the fundamental exploratory nature of reinforcement learning. Other machine learning approaches are able to improve the ability to train reinforcement learning agents by better processing input information to help map states to actions - convolutional and recurrent neural networks are helpful when input data is in image form that does not satisfy the Markov property. The specific required architecture of these convolutional and recurrent neural network models is not obvious given infinite possible permutations. There is very limited research giving clear guidance on neural network structure in a RL (reinforcement learning) context, and grid search-like approaches require too many resources and do not always find good optima. In order to address these, and other, challenges associated with traditional parameter optimization methods, an evolutionary approach similar to that taken by Dufourq and Bassett (2017) for image classification tasks was used to find the optimal model architecture when training an agent that learns to play Atari Pong. The approach found models that were able to train reinforcement learning agents faster, and with fewer parameters than that found by OpenAI’s model in Blackwell et al. (2018) - a superhuman level of performance. 2020-02-19T12:18:14Z 2020-02-19T12:18:14Z 2019 2020-02-19T12:17:41Z Master Thesis Masters MSc http://hdl.handle.net/11427/31170 eng application/pdf Department of Statistical Sciences Faculty of Science |
| spellingShingle | statistical sciences Cuningham, Blake Evolutionary algorithms for optimising reinforcement learning policy approximation |
| thesis_degree_str | Master's |
| title | Evolutionary algorithms for optimising reinforcement learning policy approximation |
| title_full | Evolutionary algorithms for optimising reinforcement learning policy approximation |
| title_fullStr | Evolutionary algorithms for optimising reinforcement learning policy approximation |
| title_full_unstemmed | Evolutionary algorithms for optimising reinforcement learning policy approximation |
| title_short | Evolutionary algorithms for optimising reinforcement learning policy approximation |
| title_sort | evolutionary algorithms for optimising reinforcement learning policy approximation |
| topic | statistical sciences |
| url | http://hdl.handle.net/11427/31170 |
| work_keys_str_mv | AT cuninghamblake evolutionaryalgorithmsforoptimisingreinforcementlearningpolicyapproximation |