A Random Field Model and its Application in Industrial Production

Julie Oger (1) , Emmanuel Lesigne (1) , Philippe Leduc (2)

1 LMPT - Laboratoire de Mathématiques et Physique Théorique
2 ST-TOURS - STMicroelectronics [Tours]

We propose a new tool of decision support in front of a globally unknown phenomenon which is modeled by a random field representing simultaneously our knowledge and our lack of information. This tool is the distribution of a random variable called failure risk probability. Before giving the precise definition of this object, we describe a context in which the decision problem appears and we discuss Bayesian random field model constructions.

Mots clés: Kriging; Bayesian inference; Gaussian processes mixture prior; multivariate t-distribution; uncertainty analysis; manufacturing yield evaluation; decision support

Domaines: Statistiques [math.ST]; Théorie [stat.TH]; Applications [stat.AP]

Julie Oger, Emmanuel Lesigne, Philippe Leduc. A Random Field Model and its Application in Industrial Production. 2015. ⟨hal-00914192v2⟩