A 1DPooling layer would be great !

NEGU93 / cvnn

Library to help implement a complex-valued neural network (cvnn) using tensorflow as back-end

https://complex-valued-neural-networks.readthedocs.io/

MIT License

164 stars 34 forks source link

A 1DPooling layer would be great ! #10

Open mhsacampos opened 3 years ago

mhsacampos commented 3 years ago

Dear Barrachina, congratulations once again for your library

It would be so useful whether the CVNN library could also have a 1DPooling layer for 1D problems that arise.

Could you provide that, please, when you have a chance ? Thanks in advance

NEGU93 commented 3 years ago

Sorry for the long time to respond. I implemented ComplexAvgPooling1D. See here the documentation. Unfortunately, Max pooling is much difficult to implement and will not be done in the near future.

For implementing max unpooling I will need tf.nn.max_pool_with_argmax to support 1D. I will issue that feature request to tensorflow.