Fulltext available Open Access
Title: Analysis and Evaluation of Reinforcement Learning Algorithms for a Continuous Control Problem
Language: English
Authors: Babic, Michael 
Keywords: Kontinuierliche Kontrolle; Agenten; Lernende Agenten; Verstärktes Lernen; Machinelles Lernen; Soft Actor Critic; Truncated Quantile Critics; Optuna; OpenAI
Issue Date: 7-Feb-2024
Abstract: 
Die große Vielfalt an an Reinforcement Learning Algorithmen macht es schwer zu bestimmen, welcher Algorithmus für welche Aufgabe verwendet werden soll. Die wissenschaftlichen Arbeiten, die solche Algorithmen präsentieren, enthalten oft wiedersprüchliche Ergebnisse und machen es dadurch noch schwerer zu verstehen, ob die Erweiterungen der grundlegenden Algorithmen eine Leistungsverbesserung aufweis...

Due to the broad variety of Reinforcement Learning algorithms, it is difficult to determine which one to use for what task. Papers that present said algorithms often claim contradictory results which worsens this problem and makes it harder to understand if their extensions of the base algorithms bring an overall improvement in performance. This work presents an approach to analyze a custom create...
URI: http://hdl.handle.net/20.500.12738/14777
Institute: Department Informatik 
Fakultät Technik und Informatik 
Type: Thesis
Thesis type: Bachelor Thesis
Advisor: Neitzke, Michael 
Referee: Becke, Martin 
Appears in Collections:Theses

Files in This Item:
Show full item record

Page view(s)

117
checked on Apr 4, 2025

Download(s)

125
checked on Apr 4, 2025

Google ScholarTM

Check

HAW Katalog

Check

Note about this record


Items in REPOSIT are protected by copyright, with all rights reserved, unless otherwise indicated.