Deep Reinforcement Learning : A Study of Reinforcement Learning with Neural Networks in Industrial Automation
Deep learning is a type of machine learning that has attracted a lot of attention in recent years because to its incredible achievements in a variety of applications like pattern recognition, audio recognition, computer vision, and natural language processing. Deep learning approaches can also be used with reinforcement learning methods to create effective representations for situations with high dimensional raw data input, according to recent study. Deep Reinforcement Learning has made it possible to learn policies for complicated tasks in partially observable settings without having to master the tasks' underlying model. Production systems face significant problems as a result of shorter product development cycles and fully customizable goods. These are required to handle not only a greater variety of products but also high throughputs, high flexibility, and resistance to process changes and unforeseen catastrophes. Deep Reinforcement Learning (RL) has been used more and more for production system optimization to address these issues. In Deep RL, recently gathered sensor-data are utilized unlike conventional Machine Learning (ML) techniques enabling real-time responses to the changes in the system. Although deep RL is now being used in production systems, it has not yet been possible to conduct a thorough analysis of the outcomes. This paper's main contribution is to give researchers and practitioners an overview of relevant applications and to inspire additional deep RL enabled production system implementations and research. The results show that deep RL is used in a range of industrial domains, supporting flexible and data-driven operations. In the majority of applications, traditional approaches performed better, requiring less effort to deploy or relying less on human expertise. However, in order to analyse safety concerns and establish reliability under real-world settings, future research needs concentrate more on applying the findings to practical systems. This article examines the applications of Deep Reinforcement Learning and its recent breakthroughs, focusing on the most commonly used deep architectures in relevance to Industrial Automation and Production Systems
Year of publication: |
2023
|
---|---|
Authors: | Iroshan, Asiri |
Publisher: |
[S.l.] : SSRN |
Subject: | Neuronale Netze | Neural networks | Lernprozess | Learning process | Lernen | Learning | Automatisierung | Automation | Theorie | Theory |
Saved in:
freely available
Extent: | 1 Online-Ressource (13 p) |
---|---|
Type of publication: | Book / Working Paper |
Language: | English |
Notes: | Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments June 26, 2022 erstellt |
Other identifiers: | 10.2139/ssrn.4386667 [DOI] |
Classification: | Y20 - Introductory Material |
Source: | ECONIS - Online Catalogue of the ZBW |
Persistent link: https://www.econbiz.de/10014259318
Saved in favorites
Similar items by subject
-
Machine learning advances for time series forecasting
Masini, Ricardo P., (2020)
-
Deep learning, predictability, and optimal portfolio returns
Babiak, Mykola, (2020)
-
Contracts for difference: a reinforcement learning approach
Zengeler, Nico, (2020)
- More ...