Compound Credibility for Conflicting Evidence Combination: An Autoencoder-K-Means Approach

Abstract

A variety of complicated problems can be solved using information fusion. Due to the inevitable existence of conflicting information in the real world, the outcome from information fusion may be inaccurate. Some methods have been proposed to date to handle conflict management. They typically consider the explicit information of the evidence, while the essential but implicit information behind the evidence is ignored to some extent. In order to overcome this issue, a novel conflicting evidence combination approach called autoencoder- <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K-</i> Means (AE- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${K}$ </tex-math></inline-formula> -Means) is proposed in this article. In AE- <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${K}$ </tex-math></inline-formula> -Means, for determining the weight of evidence, a new compound credibility has been defined, which is not only based on autoencoder, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$K$ </tex-math></inline-formula> -Means algorithm, but also evidence similarity. The compound credibility effectively integrates the explicit and implicit information of the evidence, thereby helping to further obtain dominant information and eliminate the interference from other factors. Some numerical experiments and real-world fault diagnosis examples are utilized to illustrate the efficiency of our proposed approach.