Confused about convolutional sliding window


I think that in the following image , in the center row, the convolutions operations after the max pool are no longer fully conected (compared to the first row). I mean, the green square at the end is connected to the green square before it. Am I right?

Hey @David_E1,
This is an interesting thing that you have pointed out, since a fully-connected (FC) layer by definition, connects all the neurons of the previous layer with the next layer. In the last row, I am assuming due to this reason only, FC hasn’t been mentioned. Similarly, in the center row since all the neurons are no longer connected to every other neuron, even the layers after the Max Pool layer are Conv layers. I will report this right away. Thanks again for pointing it out :blush:


Your understanding is correct.

You are welcome, and thanks for your quick response!

1 Like