Visualization method for depth model classification results facing ECG data

user-6049a1c04c775ef9a766372e(2019)

引用 0|浏览3
暂无评分
摘要
The invention discloses a visualization method for depth model classification results facing ECG data. The method comprises the following steps: inputting an ECG sequence into a trained depth model toobtain a reference result; removing the information of a selected heartbeat interval by a shield interval, comparing the output result of the depth model when the heartbeat interval information is not selected with the reference result output by the depth model, and calculating the impact factor delta O of each heartbeat for the depth model; and performing visualized expression of the impact factor delta O of each heartbeat by a gradual change color band to visualize the classification results of the depth model. By analyzing the influence of the ECG data on the output of the depth model under macro and micro granularity, the method can show the key evidence of the model classification result, and can enhance the interpretability of the classification result of the model output.
更多
查看译文
关键词
Heartbeat,Visualization,Pattern recognition,Interpretability,Interval (mathematics),Macro,Expression (mathematics),Granularity,Computer science,Artificial intelligence,Classification result
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要