How to cascade two CNN in series or parallel and combine outputs to generate single prediction in matlab?

Hello,

I wonder how can I build two or more separate customized Convolutional Neural Networks, and combine the output at the end?.. I want to feed the networks with multiple views of a single 3D image, and classify the results into one of two classes, o or 1.

One way is to average the outputs, so, can I extract the output of the fully connected layer, or the probability of the softmaxlayer to use it?

or there's another feature available to do this in matlab 2017b?

Answer 1

[André Brás]

Did you solve this question?

[sally]

not yet.. I had to work with the final output (zeros and ones) and perform logical OR or AND between them.. but I wish there was a way to get the internal output prior to the last layers