Rozvažování o cílech rozhodování: kdy sobeckost se nevyplatí?

školitel: Tatiana V. Guy
e-mail: zobrazit e-mail
typ práce: dizertační práce
zaměření: MI_MM, MI_AMSM
popis: Abstrakt: Dynamické přehodnocování cílů rozhodování. Penalizace změn cílů rozhodování závisící na ceně rozvažování nutného ke změně. Jak se vyrovnat se závislostmi mezi cíli rozhodování (zvláště rozpornými)? Jak upravit kritérium? Jak má být zohledněna cena dosažení dílčího cíle? Title: Negotiation about DM goals: when does selfishness cost less? Abstract: Dynamical reconsidering of DM goals. Penalisation of modification of DM goal in dependence on the deliberation cost of the change. How to cope with possible (inter-)dependency of DM goals (especially conflicting ones)? How to modify the criterion? How the cost of reaching the particular DM goal can be taken into account? Description: The distributed decision making (DM) arise when there is no possibility to govern in a centralised way. Any DM task of that type consider either several interacting, locally independent elements, which have their local goals, but have to collaborate to reach a common (group) goal (e.g.: cooperative robots, multi-agent systems, sensor networks); or a set of independent elements with own goals that need to coordinate their activities (e.g. vehicles in urban transportation). The success of a distributed solution depends on efficiency of cooperation. Even in the case when elements are independent they may benefit from the creation of a common DM goal - two are better than one because they can help each other succeed. If I respect my neighbour’s interests, I will reach my own DM goal with less energy. Otherwise, there is a danger the interacting elements will compete even without an explicit conflict of their DM goals. However, any negotiation costs significant time and computation resources each of interacting elements. Evaluation and optimising the tradeoff between the computational cost (time) of negotiation and a potential reward (given by a reached compromise between DM goals) should be an essential part of feasible normative DM theory. The PhD project will focus on the DM goal deliberation formulated and solved as a decision-making problem within the fully probabilistic framework [1]. The expected contribution should finally serve to normative Bayesian distributed DM under a flat cooperation structure. The preliminary conceptual solution relies on a combination of the inductive memory-based technique [3] and fully probabilistic extension of Bayesian DM [1], [5]. The proposed topic of doctoral theses can be tailored to the specific PhD student, his/her skills and inclination to theoretical or algorithmic development possibly with emphasise placed on a specific application domain.
literatura: Literature [1] Kárný M., Guy T.V.: Fully probabilistic control design. Systems & Control Letters, 55(4), 2006 [2] Kárný M., Kroupa T.: Axiomatisation of fully probabilistic design, Inf. Sciences,186, 1, 105-113, 2012 [3] Artificial Intelligence Review: Special Issue on Lazy Learning 11(1–5), 1–6, 1997. [4] Guy T.V, Kárný M, Wolpert D.H. (Eds): Decision Making with Imperfect Decision Makers, Springer, vol. 28, 2012 [5] Kárný M .,Guy T.V, Bodini A, Ruggeri F.: Cooperation via sharing of probabilistic information, Int. J. of Computational Intelligence Studies, 1:2 139-162, 2009
poznámka: Good knowledge of English is essential.
naposledy změněno: 20.03.2013 11:10:56

za obsah této stránky zodpovídá: Ľubomíra Dvořáková | naposledy změněno: 12.9.2011
Trojanova 13, 120 00 Praha 2, tel. 224 358 540, pevná linka 224 923 098, fax 234 358 643
České vysoké učení technické v Praze | Fakulta jaderná a fyzikálně inženýrská | Katedra matematiky