表 5. Comparison with existing methods.

与现有方法的对比

分类模型	整体Acc	Sen	Spe
文献[4]的 集成SVM	94.40%	65.26%	93.25%
文献[6]的CNN-LSTM	97.15%	95.14%	96.80%
文献[9]的CNN-ELM	98.51%	94.41%	98.45%
文献[10]的残差CNN	99.06%	93.21%	97.67%
本文ResNet-ABiGRU	99.12%	95.53%	99.36%