Automated heartbeat classification based on deep neural network with multiple input layers
作者：
Highlights：
• A novel deep neural network with multiple input layers is proposed.
• Automatically extracted features and hand-craft features are combined.
• CNN and LSTM are employed to extract implicit features and sequential features.
• Different strides are set for different inputs in the convolution process.
• Accuracy of 99.26% and 94.20% were achieved under class-oriented scheme and subject-oriented scheme.
摘要
•A novel deep neural network with multiple input layers is proposed.•Automatically extracted features and hand-craft features are combined.•CNN and LSTM are employed to extract implicit features and sequential features.•Different strides are set for different inputs in the convolution process.•Accuracy of 99.26% and 94.20% were achieved under class-oriented scheme and subject-oriented scheme.
论文关键词：Electrocardiogram,Heartbeat classification,Convolutional neural network,Long short-term memory,Multiple input layers
论文评审过程：Received 26 March 2019, Revised 29 July 2019, Accepted 12 September 2019, Available online 16 September 2019, Version of Record 20 January 2020.
论文官网地址：https://doi.org/10.1016/j.knosys.2019.105036
