Using Random Sets to Model Learning in Manufacturing

Paul Speaker, C. R. MacCluer


It is widely observed that  manufacturing quality metrics improve as experience is gained during production. The traditional empirical learning curves modeling such improvements have recently been explained by a predictive model deduced from first principles, namely certain principles imported into artificial intelligence from statistical mechanics. However, this new learning model  is limited to a finite lesson pool of paradigm shifts. This paper presents an extension to incremental learning using  sampling based on  the notion of dynamic random sets.


Learning curve; Manufacturing costs; Random sets; Management decision making

Full Text:



eISSN 0975-5748; pISSN 0974-875X