A Feature Reinforcement Method Based on Discrete Fourier Transform
Shengjia Zhang, Fuyuan Xiao · 2022
Complex evidence theory (CET), which extends Dempster Shafer theory to a higher dimensional space, is well studied these days. In CET, how to generate CBBA is crucial for uncertainty reasoning, and it is still an open issue. To be specific, CBBA can be generated through data in real number field or in the complex number field. For the data in real number field, it is necessary to mapping it to complex number field. Meanwhile, it is better to reinforce features of data in the process of mapping. In this paper, a discrete Fourier transform-based feature reinforcement method is proposed to map the data from real number field to complex number field. What's more, we propose a novel reinforcement measurement, similarity indices, to quantify the level of the feature reinforcement. The experiments illustrate that the discrete Fourier transform-based feature reinforcement method achieves good performance, which can be beneficial to pattern recognition to some extent.