Skip to main content
1 of 9

How are biases updated when batch size is greater than 1?

This is my network represented in matrices: (The dot represents a number) enter image description here

Feed-forwarding: enter image description here Backpropagation enter image description here

The question

enter image description here

What did go wrong?

I suppect it's because I have to intialize my bias with the shape of (neuronsInGivenLayer, batch_size) as opposed to (neuronsInGivenLayer, 1) ?