Fully probabilistic design of decision-making strategies
The Fully Probabilistic Design (FPD) framework was established in the early 1990s, extending the theory of dynamic, expected-utility-based decision making (DM). This innovative framework has been continuously evolving and refining, which further enhances its relevance and applicability in various decision-making contexts.
FPD of decision rules (strategies, policies, controllers) [1] is an axiomatic strict extension of standard Bayesian decision making.
This theory models the agent’s environment and decision goals through probabilities, extending existing theories based on the optimization of expected utility [1]. It probabilistically models not only the behaviour of the decision loop but also its desired Ideal.
FPD enhances tools as the minimum relative entropy principle [2] or Bayes’ rule [3], improves feasible adaptive certainty-equivalence-based DM [4] and many others, e.g. [5].
Contact: Miroslav Karny
[1] Kárný M.: Axiomatisation of Fully Probabilistic Design Revisited, Systems and Control Letters 141, 2020
[2] Kárný M., Guy T.V.: On Support of Imperfect Bayesian Participants, Decision Making with Imperfect Decision Makers, 29-56, 2012
[3] Quinn A., Kárný M., Guy T.V.: Fully probabilistic design of hierarchical Bayesian models, Information Sciences 369(1): 532-547, 2016
[4] Kárný M.: Fully probabilistic design of strategies with estimator, Automatica, 2022
[5] Kuklišová P. L., Jirsa L., Quinn A.: Fully probabilistic design for knowledge fusion between Bayesian filters under uniform disturbances, Knowledge-Based System, 238, 107879, 2022