HW question about hidden layer parameters

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi Mentors and friends,

Suppose your input is a 300 by 300 color (RGB) image, and you use a convolutional layer with 100 filters that are each 5x5. How many parameters does this hidden layer have (including the bias parameters)?

The answer is (5x5+1)x100 = 2600 Is this correct ? I do already pass the hw, but this is one I didn’t get right. THanks.

2

No, that’s not correct. Don’t forget that each filter also has a third dimension to account for the 3 input channels, right?

3

wow… so, I should do 5sz5sz3ch*100filters+100bias = 7600 ?

4

Yes, each filter will have 5 x 5 x 3 + 1 parameters that are trainable.