An ST segment classification neural network of a high-order polynomial activation function. Removal of random noise by means of mean filtering has a good effect, a baseline is removed by means of wavelet filtering, and noise in a signal is finally removed. Moreover, on the basis of a combination of a convolutional neural network and a high-order polynomial activation function, the complexity of a model can be directly increased by means of great divergence of the high-order polynomial function, the problem of hyper-parameter selection in a regularization process is avoided, and thus the generalization capability of the neural network is significantly improved.
展开▼
