Reshape layer in deep learning

I want to try out a deep learning architecture which first does fully connected layers and then transitions into image convolutions. So after the fully connected layers I want to reshape the weights from a vector into a image like matrix. Is there a reshape layer within the Deep learning toolbox which does this?

Answer 1

[Shivam Sardana]

There is no reshape layer in MATLAB which changes output from fully connected layer into image like matrix. You can create a custom deep learning layer to solve this problem. You can refer to this documentation link for more details on creating a custom layer: https://www.mathworks.com/help/deeplearning/ug/define-custom-deep-learning-layers.html

Hope this helps.

Answer 2

[Tomaso Cetto]

Hi KB,

As of R2022a, the Deep Learning Toolbox has a new functionLayer - this layer enanbles you to apply a function to the activations of your network. Using this layer is the preferred way of performing the reshaping operation you're after. The link above has an example of how to do this, under the 'Reformat Data Using Function Layer' heading.

Hope this helps!