[lec content] Convolution Neural Network matrix shape, 4 x 4 x 4 error?

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi teaching team, just want to make sure if I understood this correctly.
In the lecture (week1 one-layer-of-a-convolutional-network ) delivered by Dr. Andrew Ng,

around time 2:05 he mentioned that one layer of CNN is the layer that turns 6 x 6 x 3 input to 4 x 4 x 4 output ?
Did he mean 4 x 4 x 2, instead ? because on the slide, he wrote down 4 x 4 x 2 ?

Thanks

image
image1152×720 101 KB

2

I think you’re right that he meant 4 x 4 x 2. The point is that each filter gives one channel of the output, so the number of channels in the output is equal to the number of individual filters that are applied. In the example he is showing, there are 2 filters, so the output shape from that layer will be 4 x 4 x 2.

3

Thanks