Complex Belief Divergence Measures for Multisource Information Fusion
针对多模态数据源中的模糊信息，在复杂证据理论框架下提出两种新的散度度量（CBKL和CBJS），用于量化证据体间的差异，并基于此设计多源信息融合方法，通过分类任务验证了其在复杂证据融合与决策中的有效性。
Uncertainty is a crucial aspect in real-life scenarios, especially when dealing with ambiguous information from multimodal data sources. Complex evidence theory (CET), a generalized form of Dempster-Shafer evidence theory that uses complex numbers to describe uncertainty, offers a more promising framework for dealing with uncertainty in many fields. Focusing on conflict management in the CET framework, new divergence measures, namely, complex belief Kullback-Leibler (CBKL) divergence and complex belief Jensen-Shannon (CBJS) divergence, are proposed in this article to quantify the discrepancy between complex evidence bodies. Additionally, novel multisource information fusion methods based on CBKL divergence and CBJS divergence for decision making are proposed. To validate the effectiveness of the proposed methods, classification tasks are performed. The results demonstrate the feasibility and accuracy of the proposed methods in handling complex evidence fusion and decision-making tasks.