Abstract
AbstractIn this paper, the landscape framework is used to analysis the tracking performance of univariate marginal distribution algorithm (UMDA) in dynamic environment. A set of stochastic differential equations (SDEs) is used to describe the evolutionary dynamics of the algorithm. The corresponding potential function is constructed from these SDEs. Dynamic mean first passage time, which is a new concept, is defined as the time it takes from an optimum to another in a dynamic environment. This concept can be used to measure the tracking property of the algorithm.KeywordsUMDADynamic environmentPotential functionDynamic mean first passage time
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have