Partially Observable Risk-Sensitive Markov Decision Processes

DOI10.1287/MOOR.2016.0844MaRDI QIDQ4595962zbMATH OpenOpenAlexFDO

Authors Nicole Bäauerle, Ulrich Rieder

Publication date 7 December 2017

Published in Mathematics of Operations Research (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1504.03530

zbMATH Keywords

value iterationexponential utilitycertainty equivalentupdating operatorpartially observable Markov decision problem

Mathematics Subject Classification ID

Markov and semi-Markov decision processes (90C40) Stochastic learning and adaptive control (93E35)

Abstract: We consider the problem of minimizing a certainty equivalent of the total or discounted cost over a finite and an infinite time horizon which is generated by a Partially Observable Markov Decision Process (POMDP). The certainty equivalent is defined by $U^{-1}(EU(Y))$ where $U$ is an increasing function. In contrast to a risk-neutral decision maker this optimization criterion takes the variability of the cost into account. It contains as a special case the classical risk-sensitive optimization criterion with an exponential utility. We show that this optimization problem can be solved by embedding the problem into a completely observable Markov Decision Process with extended state space and give conditions under which an optimal policy exists. The state space has to be extended by the joint conditional distribution of current unobserved state and accumulated cost. In case of an exponential utility, the problem simplifies considerably and we rediscover what in previous literature has been named information state. However, since we do not use any change of measure techniques here, our approach is simpler. A small numerical example, namely the classical repeated casino game with unknown success probability is considered to illustrate the influence of the certainty equivalent and its parameters.

Recommendations

• Risk-sensitive finite-horizon piecewise deterministic Markov decision processes
• On risk-sensitive piecewise deterministic Markov decision processes
• Risk-Constrained Markov Decision Processes
• More risk-sensitive Markov decision processes
• Partially observable Markov decision processes with imprecise parameters
• Risk-averse dynamic programming for Markov decision processes
• Stochastic Approximation for Risk-Aware Markov Decision Processes
• Partially observable Markov decision processes incorporating
• Markov decision processes with recursive risk measures
• Continuous-time Markov decision processes with risk-sensitive finite-horizon cost criterion

Cited in

(16)

• Minimal disclosure in partially observable Markov decision processes
• Finite horizon partially observable semi-Markov decision processes under risk probability criteria
• Partially observable discrete-time discounted Markov games with general utility
• Time-inconsistent risk-sensitive equilibrium for countable-stated Markov decision processes
• Zero-sum risk-sensitive stochastic games
• Markov decision processes under ambiguity
• Risk-averse dynamic programming for Markov decision processes
• Risk-sensitivity vanishing limit for controlled Markov processes
• Risk measurement and risk-averse control of partially observable discrete-time Markov systems
• Optimal control of partially observable piecewise deterministic Markov processes
• On risk-sensitive piecewise deterministic Markov decision processes
• Partially observable Markov decision process approximations for adaptive sensing
• Markov risk mappings and risk-sensitive optimal prediction
• Markov decision processes with risk-sensitive criteria: an overview
• Partially observable discrete-time stochastic games under risk probability criterion
• Risk-sensitive discounted Markov decision processes with unbounded reward functions and Borel spaces

This page was built for publication: Partially Observable Risk-Sensitive Markov Decision Processes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4595962)