A New Method to Determine Generalized Basic Probability Assignment in the Open World

Abstract

Dempster-Shafer evidence theory (D-S theory) is a very useful tool to solve problems in the field of information fusion. But how to determine generalized basic probability assignment (GBPA) more accurately and efficiently in D-S theory is still a matter of debate, especially in the open world. In this paper, we put forward a new method to determine GBPA in the open world. First, a minimum spanning tree (MST) is established for samples in each known class. The covering radius of every edge in MST of each class will be generated based on the formula for generating improved radius. The MST coverage of every class will be established. Combine these coverages to form the MST covering model. Each new sample should be justified the MST coverages it belongs to. Then, we put forward a formula to generate GBPA of the sample. Finally, determine whether the sum of GBPA is smaller than 1. If so, m(Ø) needs to be generated. Otherwise, GBPA needs to be standardized. The experimental results on Iris dataset prove the effectiveness of our method.

Keywords

• Dempster-Shafer evidence theory
• generalized evidence theory
• basic probability assignment
• open world
• minimum spanning tree covering model