0
$\begingroup$

So I got an EEG dataset with shape (data points, 19), each row's shape (1,19) represent 1 second of EEG.

I read much research on EEG classification that used many Deep Learning method and 1D-CNN is one of that.

My question is as the input of the 1D-CNN must have multi-row data, ex (50,19) for my dataset so it can filter a input matrix. But I want to predict new data row by row ((1x19) shape), can 1D-CNN use this input for predict new data?

$\endgroup$
1
$\begingroup$

The inputs of a CNN must have the same shape during prediction as when it was trained. So if you have a CNN trained on 50 time-steps windows, then you can make predictions on a stream of input by updating the data in this window continuously. Each new time step you push the most recent row onto the end, and drop the earliest row.

Of course it is possible to make a CNN model that take 1x19 as the input also. But such a model would not have any ability to detect patterns over time, which would probably limit its predictive power quite a lot.

$\endgroup$
1
  • 1
    $\begingroup$ Thank you very much. That help me alot. $\endgroup$
    – Q.H.Chu
    Jun 29 '20 at 2:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.