基于对比学习和多尺度双维注意力机制的心电分类方法
By using the MFE-ResNet network model and combining it with a multi-scale two-dimensional attention mechanism, the problem of insufficient feature correlation modeling in ECG signal classification by traditional models is solved, achieving higher feature capture ability and classification accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES)
- Filing Date
- 2023-11-16
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional deep learning models fail to adequately model the correlation of features across channels and time in ECG signal classification, resulting in poor classification accuracy.
An ECG classification method based on contrastive learning and multi-scale two-dimensional attention mechanism is adopted. The method uses the MFE-ResNet network model, combined with parallel convolution module, Batch_Norm layer, ReLU activation function layer, residual and dilated convolution module, multi-scale two-dimensional attention module and feature fusion module to fully model the correlation of features in channels and time.
It improves the ability to capture ECG signal features and the accuracy of classification, and enhances the model's generalization ability and robustness.
Smart Images
Figure CN117653137B_ABST