1. Digital Information Feature Compression Method Based on Weighted Trust Vector
- Author
-
Zhu Li, Wan Yi, Xiao Kun, Guo Shuying, and Deng Tao
- Subjects
Computer science ,business.industry ,Feature (computer vision) ,Encoding (memory) ,Compression (functional analysis) ,Key (cryptography) ,Pattern recognition ,Artificial intelligence ,Cluster analysis ,business ,Convolutional neural network ,Data modeling ,Coding (social sciences) - Abstract
In order to improve the digital information feature compression ability of multi-domain convolutional neural network. The formula coding model of multi-domain convolutional neural network digital information feature compression is constructed, the feature coding in the process of multi-domain convolutional neural network digital information feature compression is combined with the randomness of weighted trust vector, the information fusion process is combined with the randomness of weighted trust vector, the data feature analysis model of multi-domain convolutional neural network digital information feature compression is established, and the data coding result is obtained by deep fusion algorithm for feature information coding reconstruction. The weighted trust feature quantity of digital information coding and feature compression is established. Under the control of cyclic shift key optimization, the multi-domain convolutional neural network digital information feature recombination is obtained by differential evolution clustering and information fusion, and the digital information feature compression is realized. Simulation results show that the output stability of digital information feature compression by this method is good, the security of digital information transmission is high, the linear spatial combination feature expression ability of digital information is strong, and the anti-interference ability is strong.
- Published
- 2020