How to do reshape in a NN?

How can I reshape a NN's layer's output? I have a fully connected layer whose output is a 1D vector. I want to reshape it to a 3D matrix and apply convolution layer after it. Say, is there a similar function or something like "reshape layer" in Tensorflow?

Answer 1

[Rahul]

Hi @Ma Yuelin,

There are multiple ways to approach the issue in MATLAB. The following can be considered:

• 'Deep Network Designer' app can be used to adjust layers according to custom requirement.
• Since there is no direct 'reshapeLayer' function avaialble, a similar functionality can be acheived using 'funcionLayer' function. This is also mentioned in this MATLAB Answer: https://www.mathworks.com/matlabcentral/answers/461988-reshape-layer-in-deep-learning. The section of 'Reformat Data Using Function Layer' would provide necessary information.
• Defining a custom class for the output layer, where the 'reshape' function can be used to adjust the shape of the output layer.

The following Mathworks documentations can be referred to know more:

'Deep Network Designer': https://www.mathworks.com/help/deeplearning/ref/deepnetworkdesigner-app.html

'functionLayer': https://www.mathworks.com/help/deeplearning/ref/nnet.cnn.layer.functionlayer.html

'reshape': https://www.mathworks.com/help/matlab/ref/double.reshape.html

'Defining custom Deep Learning Layer': https://www.mathworks.com/help/deeplearning/ug/define-custom-deep-learning-layer-with-formatted-inputs.html

Thanks.