Full text
| |
| Author(s): |
Daniel Kikuti
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Escola Politécnica (EP/BC) |
| Defense date: | 2008-08-20 |
| Examining board members: |
Fabio Gagliardi Cozman;
Leliane Nunes de Barros;
Newton Maruyama;
Leandro Chaves Rêgo;
José Carlos Ferreira da Rocha
|
| Advisor: | Fabio Gagliardi Cozman |
| Abstract | |
In this thesis we explore situations where preferences are partially ordered: given two acts, the agent may prefer one to another, or nd them to be equivalent, or nd them to be incomparable. Such preferences stem from the uncertainty associated to some states of the decisions model and are revealed by imprecision in probability values. We investigate six criteria for strategy selection in decision trees and inuence diagrams with imprecise probabilities: -maximin, -maximax, -maximix, Interval Dominance, Maximality and E-admissibility. We present new algorithms that generate strategies for all these criteria. The main contributions of this work are twofold: the implementation of these algorithms and the analysis, under the computational point of view, of the criteria considered ratio- nal in uncertain situations represented by set of probabilities. (AU) | |