Matrix Multiplication or regular

@jeremy The layers_example.xls from lesson 2 shows matrix multiplication but the conv_example.xls in class 4 did a regular multiplication of all elements of the matrix and added it up.

If I understand you right -
the layers spreadsheet is showing how a dense layer works, not a convolution.
Convolution will be done with regular multiplication and dense layer works with matrix multiplications. Correct?

1 Like

Correct :slight_smile: A convolution does element-wise multiplication of each filter with it’s input window, then adds them all up.

(Note that the far right of conv_example also shows a dense layer - but because there’s only one output activation for that layer, a matrix multiplication is simply a sumproduct.)

1 Like