Week 1 Assignment 1 Convolution Model Conclusion


On the “What you should remember” section after exercise 4 pool forward, it is stated that:
“A convolution extracts features from an input image by taking the dot product between the input data and a 3D array of weights (the filter).”

Shouldn’t a convolution is an element-wise multiplication between input and filter?

Hi Stelis,

Thanks for your interesting question about which there is some debate. You can have a look here for clarification. Happy reading!

Thanks for the reference.