Explanation why NN don't work on images and why we need CNNs

[EdwinB12]

I think the Convolutional Neural Network section should start with a explanation of why flattening the input (and therefore the NN approach that was done in Lab 02) is not ideal for images.

[MaxGamill-Sheffield]

@EdwinB12 Ah yes I forgot to address this in my comment.

I was thinking of moving it to either:

1. Second to last - before the "put it all together" slide or...
2. After "stacking convolution layers" - before mentioning the pooling layers

But I was curious if @twinkarma wanted to make this point more general, and also applied to deep dense networks (hence being at the start) and solely with a focus on CNN's?

What do you both think?