A New Method to Determine Basic Probability Assignment Based on Interval Number
Siqiang Zhong, Xingcheng Liu · 2019
Dempster-Shafer evidence theory(D-S theory) is a widely used method in information fusion fields. The significant step in applying D-S theory is to determine the basic probability assignment(BPA). But how to determine BPA appropriately is still a problem to be solved. In this paper, we propose a new formula for calculating the similarity of interval numbers and introduce attribute weights based on the uncertainty of different attributes. Then, the BPA is determined by using the attribute weights, and the similarity between the interval number models and the test samples. Finally, some classification experiment of UCI dataset is used to illustrate the validity of this method.