Multi-label multi-mode holographic pulse condition recognition method based on graph convolution network

• Main Authors: ZHANG LIJIA, QIN JIANZENG
• Format: Patent
• Language: Chinese; English
• Published: 12.06.2020
• Subjects: CALCULATING; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATIONTECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING ORPROCESSING OF MEDICAL OR HEALTHCARE DATA; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a multi-label multi-mode holographic pulse condition recognition method based on a graph convolution network. A relation matrix is constructed by adopting a data driving mode;a graph neural network is adopted to mine co-occurrence modes of pulse condition tags and tags and tags and data in a data set to define correlations between the tags and between the tags and the data; then, the features of pulse condition video are extracted by adopting space-time separable 3D convolution; the whole model structure adopts 2D convolution operation at the front; the space-time separable 3D convolution operation is carried out at the back, and finally data fusion is carried out in a weighted point multiplication mode according to the pulse condition video feature vector and thepulse condition relationship feature vector extracted by the space-time separable 3D convolution and graph neural network, so that the pulse diagnosis process of the machine becomes more efficient andaccurate. 本发明公开了一种基于图卷积网络的