Let's assume I use convolutional networks for time-series prediction. Data I feed to the network have 1 channel depth, height of number of periods and number of features is the width, so the frame size is: [1, periods, features]. Batch size is not relevant here.

Is there a difference between using 1d convolutions along time (height) dimension and 2d convolutional that will have a kernel size of for example (3, 1) or (5, 1), so that the larger number convolutes along the time dimension, and there is no convolution along features dimension?


Your Answer

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

Browse other questions tagged or ask your own question.