RTFN-MSIF: A Reinforced TFN-based Multisource Information Fusion for Pattern Classification

As a generalized model of Dempster–Shafer’s evidence theory, complex evidence theory (CET) has the ability to express uncertainty and reasoning uncertainty. One of the key problems in CET is how to model uncertain information in CET by means of complex basic belief assignment (CBBA). Therefore, this article proposes an uncertainty modeling method to generate CBBA using the reinforced triangular fuzzy number (TFN) function. The proposed RTFN-nps-based uncertainty method can extract feature information from the data source by taking advantage of the decomposition of complex numbers. In addition, based on the proposed CBBA generation method, a reinforced TFN-based multisource information fusion (RTFN-MSIF) method is designed for decision-making. Finally, the RTFN-MSIF method is applied to the pattern classification to verify its effectiveness. In conclusion, the RTFN-MSIF method can deal with uncertainty modeling and reasoning in both real and complex domains, and provides a promising method for decision theory.