We deal with preference learning from pairwise comparisons, in case of decision under uncertainty, using a new rough set model based on stochastic dominance applied to a pairwise comparison table. For the sake of simplicity we consider the case of traditional additive probability distribution over the set of states of the world; however, the model is rich enough to handle non-additive probability distributions, and even qualitative ordinal distributions. The rough set approach leads to a representation of decision maker’s preferences under uncertainty in terms of "if..., then..." decision rules induced from rough approximations of sets of exemplary decisions. An example of such decision rule is "if act a is at least strongly preferred to act a′ with probability at least 30%, and a is at least weakly preferred to act a′ with probability at least 60%, then act a is at least as good as act a′."
|Title of host publication||Computational intelligence for knowledge-based systems design: 13th international conference on information processing and management of uncertainty|
|Editors||E. Hullermeier, R. Kruse, F. Hoffmann|
|Place of Publication||Berlin|
|Number of pages||11|
|Publication status||Published - Jul 2010|
|Name||Lecture notes in computer science|