For anyone interested in the problems associated with excluding bias (b) from an equation, there are some great explorations on the subject at the below-listed links.
Overall, if I understand this correctly, each node has it’s own bias and that is fed back through the layers as the algorithm gets developed.
Links: