Högskolan i Skövde

his.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Evaluating hyperparameter optimization on the generalization of deep reinforcement learning algorithms
University of Skövde, School of Informatics.
University of Skövde, School of Informatics.
2025 (English)Independent thesis Basic level (degree of Bachelor), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Deep Reinforcement Learning (DRL) is a branch of Artificial Intelligence (AI) focused on developing decision-making systems that learn through interaction with their environment. A central challenge in DRL is generalization—the ability of trained models to perform well in previously unseen environments. This study aims to evaluate the impact of hyperparameter optimization (HPO) on the generalization capabilities of two popular DRL algorithms: Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC). HPO, typically used to improve task-specific performance by finding the optimal hyperparameters (HPs), is investigated here as a potential method to enhance generalization. Experiments were conducted to compare the performance of PPO and SAC with and without HPO across varied environments. Results indicate that SAC benefits from HPO, whereas PPO performs better with default settings. These findings suggest that the effectiveness of HP tuning in DRL is highly context-dependent, influenced by both the choice of algorithm and the characteristics of the environment.

Place, publisher, year, edition, pages
2025. , p. 44
Keywords [en]
hyperparameter optimization, generalization, deep reinforcement learning, proximal policy optimization, soft actor-critic
National Category
Computer Sciences
Identifiers
URN: urn:nbn:se:his:diva-25549OAI: oai:DiVA.org:his-25549DiVA, id: diva2:1985258
Subject / course
Informationsteknologi
Educational program
Computer Science - Specialization in Systems Development
Supervisors
Examiners
Available from: 2025-07-23 Created: 2025-07-23 Last updated: 2025-09-29Bibliographically approved

Open Access in DiVA

fulltext(4391 kB)665 downloads
File information
File name FULLTEXT01.pdfFile size 4391 kBChecksum SHA-512
40c4d1706a6f2d166ae299a675fe71014af2ab5891fe98c96d73c379a169234841d5b890aea89d7d6a9dd80c1b87838f917a7e2ff68c624eb988868dfbb6b939
Type fulltextMimetype application/pdf

By organisation
School of Informatics
Computer Sciences

Search outside of DiVA

GoogleGoogle Scholar
Total: 665 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 291 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • apa-cv
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf