Interval Markov Decision Processes with Multiple Objectives

• DOI: 10.1145/3309683
• Publication Date: 2019-11-18T13:01:53Z
• Authors: Ernst Moritz Hahn; Vahid Hashemi; Holger Hermanns; Morteza Lahijanian; Andrea Turrini
• Keywords: 0102 computer and information sciences; 01 natural sciences

Accurate Modelling of a real-world system with probabilistic behaviour is difficult task. Sensor noise and statistical estimations, among other imprecisions, make the exact probability values impossible to obtain. In this article, we consider Interval Markov decision processes ( IMDP s), which generalise classical MDP s by having interval-valued transition probabilities. They provide powerful modelling tool for systems an additional variation or uncertainty that prevents knowledge We investigate problem robust multi-objective synthesis Pareto curve analysis queries on s. study how find (randomised) strategy satisfies multiple objectives involving rewards, reachability, more general ω-regular properties against all possible resolutions uncertainties, as well generate approximate providing explicit view trade-offs between objectives. show PSPACE -hard value iteration-based algorithm set achievable points. finally demonstrate practical effectiveness our proposed approaches applying them several case studies using prototype tool.