Real processes result from interactions of relatively independently deciding but mutually interacting parts, modelled predominantly as multi-agent systems. A sufficiently complete, normative and algorithmically implementable theory is still missing, which is dearly paid for by effort and quality when solving specific problems. Preliminary results confirm that such a theory can be created by combining Bayesian decision-making, dynamic probabilistic modelling and fully probabilistic design of decision strategies. This is a still-new research area, which is widely applicable, for instance, in technology or e-democracy. It faces a range of theoretical, optimisation, algorithmic, experimental, as well as software problems suitable for 2-3 PhD. theses and a range of MSc. and BSc. theses. Their content can be tailored to the student’s interests, from purely theoretical to software or a specific application-oriented level.
References (samples to be tailored and complemented)
[1] Peterka V.: Bayesian System Identification, in P. Eykhoff “Trends and Progress in System Identification”, Pergamon Press, Oxford, 239-304, 1981.
[2] Kárný M., Guy T. V.: On dynamic decision-making scenarios with multiple participants, Multiple Participant Decision Making, 17-28, 2004.
[3] Kárný M., Guy T. V., Bodini A., Ruggeri F.: Cooperation via sharing of probabilistic information, International Journal of Computational Intelligence Studies, 139-162, 2009
[4] Sečkárová V., Hrabák P.: Optimizing Movement of Cooperating Pedestrians by Exploiting Floor-Field Model and Markov Decision Process , Bayesian Statistics in Action, 241-251, Eds: Argiento R., et al: Bayesian Young Statisticians Meeting, BAYSM 2016, 2017
[5] Kárný M., Alizadeh Z.: Towards Fully Probabilistic Cooperative Decision Making, Multi-Agent Systems: 16th European Conference, EUMAS 2018, Revised Selected, 2019
