Disjoint Programming in Computational Decision Analysis

• This paper discusses a series of imprecise decision models and their corresponding computational aspects arising in computational decision analysis. The imprecise decision models relax the traditional point estimates into intervals and incorporate various types of vague information represented as linear constraints from a decision-maker. When the principle of maximizing expected utility is applied as the decision rule, the evaluations of these models become nonconvex optimization problems and require some global optimization strategies. This paper presents a class of global optimization algorithms for solving such non-convex programs. We take advantage of polar cuts and the disjoint structural property of the imprecise decision models to develop generalized cutting plane methods that are different from the traditional class of branch and bound approaches.

Ämnesord

NATURVETENSKAP  -- Data- och informationsvetenskap -- Systemvetenskap, informationssystem och informatik (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Information Systems (hsv//eng)

Nyckelord

global optimization

disjoint programming

cutting plane

imprecise decision analysis

Computer and systems science

Data- och systemvetenskap

data- och systemvetenskap

Computer and Systems Sciences

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)