at 8:36 in the video in generator , at last a convolution layer is applied that reduced the number of channels, but what i remember is that convolution layers increase the number of channels.Please if you can clarify the reason?
Doubt in video in generator architecture
Hey @saksham_thukral, it is not necessary that convolution layer increases the number of channels. The channels will increase or decrease completely depends on the no of filters/units that are used in a single convolutional layer.
You are confusing it with the general notion that in CNNs as we move from the input layer to the output layer, there is an increase in the number of channels. That is why we sometimes say that Convolutional layer increases the number of channels. But you would observe that only in the case when the corresponding convolutional layers contain filters in the order :
Input --> 128 --> 256 --> 512 --> 1024 --> Dense layers follow --> Output layer
But if we place the convolutional layers in the order
Input --> 1024 --> 512 --> --> 256 --> 128 --> 64 --> Dense layers follow --> Output layer
Then in this case we can say that convolutional layers are decreasing the number of channels.
Hope you understand.
Happy coding 
I hope I’ve cleared your doubt. I ask you to please rate your experience here
Your feedback is very important. It helps us improve our platform and hence provide you
the learning experience you deserve.
On the off chance, you still have some questions or not find the answers satisfactory, you may reopen
the doubt.