DocumentCode
1562146
Title
Using k-means for clustering in complex automotive production systems to support a Q-learning-system
Author
Doring, Andre ; Dangelmaier, Wilhelm ; Danne, Christoph
Author_Institution
Univ. of Paderborn, Paderborn
fYear
2007
Firstpage
487
Lastpage
497
Abstract
This work shows the application of k-means clustering to reduce the state space complexity for a q-leaning algorithm in supply networks of serial production systems. An adequate clustering function is introduced and based on several scenarios the results of the clustering are validated with respect to their usability for the q-learning system. In addition, runtime and scalability aspects are evaluated for those scenarios.
Keywords
automobile industry; learning (artificial intelligence); pattern clustering; production engineering computing; state-space methods; supply chain management; adequate clustering function; complex automotive production systems; k-means clustering; machine learning; q-learning-system; reinforcement learning; state space complexity; supply networks; Artificial intelligence; Automotive engineering; Clustering algorithms; Computer networks; Humans; Information systems; Production planning; Production systems; State-space methods; Supply chain management;
fLanguage
English
Publisher
ieee
Conference_Titel
Cognitive Informatics, 6th IEEE International Conference on
Conference_Location
Lake Tahoo, CA
Print_ISBN
9781-4244-1327-0
Electronic_ISBN
978-1-4244-1328-7
Type
conf
DOI
10.1109/COGINF.2007.4341928
Filename
4341928
Link To Document