Is there any way to use "activations" function for a dlnetwork ?

I would like to obtain an activation from a specific layer of a trained network of type 'dlnetwork' by using the 'activations' function, but it does not work resulting in the following error message.

"undefined function 'activations' for input argument of type 'dlnetwork' ."

Can I get around this problem ?

Answer 1

[Sourav Bairagya]

The "activations" function supports only SeriesNetwork object or DAGNetwork object, but doesn't support 'dlnetwork' object.

However, you can use "forward" fucntion to compute activations from a specific layer of a 'dlnetwork' object.

[dlY1,..,dlYN] = forward(dlnet,dlX,'Outputs',layerNames)

You can leverage this link to know more details about the usage of "forward" function:

https://www.mathworks.com/help/deeplearning/ref/dlnetwork.forward.html?s_tid=doc_ta#function_forward_sep_mw_3ffb42b3-6af7-4c22-82b3-4fb0f6a399f2

[hyeonjin kim]

Dear Sourav,

I greatly appreciate your help !!!

Thanks very much !!!