|
|
 |
|
ECS Lecture by Florentin Smarandache |
|
Date: November 5, 2004
Time: 2:00pm
Location: NIA, Rm 404
Speaker: Florentin Smarandache of University of New Mexico
Subject: * "An In-Depth Look at Information Fusion Rules & Unification of Fusion Theories"
A glossary of about 32 fusion rules are presented or cited, among them new ones just introduced, together with counter-examples to them. Most rules first use the conjunctive rule, then transfer the conflicting mass to other (especially non-empty) sets. Each rule works better in some applications and less in others. More rules are normally required for solving a problem in dynamic fusion. Since no rule or fusion theory completely works for all applications, the author proposes a Unification of Fusion Theories and rules (UTF):
- extend the power / hyper-power sets from previous theories to a Boolean algebra (closures of the frame of discernment under union, intersection, and complement of sets; for non-exclusive elements one considers a fuzzy or neutrosophic complement – a function not necessarily involutive);
- for each application, one selects the most appropriate model, ensemble of rules, and algorithm of implementation. The UFT Scenario looks like a logical chart or like a cooking recipe. Default theory of UFT, i.e. when we don’t have extra-information on sources or on hypotheses of the frame of discernment, is DSmT, upon the average minimum principle.
Information Fusion, which is an alternative to Dempster-Shafer Theory of Evidence and an extension of Dubois-Prade Theory. DSmT takes into consideration any kind of model (free, hybrid DSm models, Shafer’s model) according to the integrity constraints of the fusion problem, and proposes a new mathematical frame work for information fusion that allows intersections of elements (degree of consensus between elements), can manage uncertainty, highly conflicting information, imprecise sources, and overcomes limitations of Dempster’s rule. DSmT is adapted to static or dynamic fusion applications represented in terms of belief functions based on the general formalism.
|
|
|
Home