5

I'm a student and a beginner on Physionet. What is the best method to classify EEG and ECG signals: KNN or SVM? And how can I do it with matlab? Should I convert the .dat files to .mat or just extract features to build my matrix for training and for testing? I don't know how to begin to develop my classifier?

philshem
  • 17,647
  • 7
  • 68
  • 170
tayssir fekih
  • 59
  • 3
  • 1
    I think current state-of-the-art for EEG datasets of seizure classification is convolutional neural networks: http://yann.lecun.com/exdb/publis/pdf/mirowski-mlsp-08.pdf – Tomas Feb 14 '17 at 19:00
  • Hi thank you for the answer, and is it the same thing for the ECG signals? the convolutional neural network, we can implement int with matlab?? means I can use the same algorithms used to extract features and to build my dataset vectors ?? thank you very much – tayssir fekih Feb 16 '17 at 10:08

1 Answers1

0

Have you tried using a bidirectional LSTM?

model = Sequential()
model.add(InputLayer(input_shape=(1, 12)))
model.add(Bidirectional(LSTM(100, activation='tanh', return_sequences=True))
bjornsing
  • 101
  • 2