Statistics Seminars: Efficient computation of system reliability under severe uncertainty
4 November 2013 14:00 in CM221
Computing efficiently the reliability bounds of big systems is already an important issue when the uncertainty of each component is characterized by a precise probability. This issue becomes even more critical when component uncertainties are characterized by lower and upper probability bounds. In this talk, we will review some recent advances concerning the efficient computation of system reliability when component uncertainty is described by a belief function, i.e., a completely monotone choquet capacity. Such uncertainty models, while not being the most generic of the literature, already encompass a number of common imprecise probabilistic models, such as p-boxes or possibility distributions. We will show that the notion of minimal cuts and paths can also be exploited for such models, both for binary and multi-state systems. If times allow, we will briefly discuss some possible extensions.
Contact email@example.com for more information