Determination of Evidence Weights Based on Convolutional Neural Network for Classification Problem

The weighted average is an efficient way to address conflicting evidence combination in the Dempster-Shafer theory. However, it open issue how determine weights reasonably. Although many traditional solutions based on distance or entropy have been put forward, are determined with a single aspect, and no comprehensive consideration of other useful information affects weights. Thus, does not ensure that determination most reasonable. By introducing deep learning into combination, this paper proposes method for determining convolutional neural network. Taking as network input corresponding weight output, utilizes fully mine potentially weights, order comprehensively. Additionally, we define loss function. continuously optimized through back propagation achieve optimal when function value minimum. Classification experimental results demonstrate proposed outperforms ones can be flexibly extended application fields decision-making fusion method.