Evidence approximation algorithm based on focal element similarity
Jun Xie, WU Gen-xiu · 2012
The computational complexity of traditional Dempster-Shafer combination rule will be increased dramatically when the number of focal element has increased, which limits the rule widely being applied. From the point of reducing the number of focal element, a kind of improved algorithm is proposed to solve the problem of computational complexity. This paper is based on the average mass value. Standard deviation was viewed as a fluctuation of the mass value, then a threshold was set up. When the basic probability assignment is less than the threshold, the corresponding focal element will be deleted. The basic probability assignment of deleted focal elements will be reassigned to the retaining focal elements. Finally, three examples shows that the improved algorithm is effective and valid by comparing with other approximation algorithms.