Feb 26 · 1 min

Its clear from your explanation that the 1st layer detects the basic features while the 2nd layer…

The first conv layer detects basic features and the second combines them in the form of a weighted sum (as per convolution). The third layer will combine features of the second layer, but this time it happens by first applying the matrix transformation to the incoming vector and then calculating the weighted sum. The matrices are learned using back…

29

1 response