Multiobjective probabilistic mixture control

Authors:	Guy Tatiana Valentine, Institute Information Theory and Automation, Czech Republic Bohm Josef, Institute Information Theory and Automation, Czech Republic Karny Miroslav, Institute Information Theory and Automation, Czech Republic
Topic:	1.2 Adaptive and Learning Systems
Session:	Switching and Multiple Model Approaches to Adaptation
Keywords:	Multiobjective control, probabilistic models, decision making

Abstract

Paper formulates the problem of multiobjective probabilisticmixture control design and proposes its general solution with bothsystem model and target represented by finite probabilisticmixtures. A complete feasible algorithmic solution for mixtureswith components formed by normal auto-regression models withexternal variable is provided .