Why does derivatives for w_j in gradient descent differ from b?

Course Q&A Machine Learning Specialization Supervised ML: Regression and Classification
1

In gradient descent both for linear and logistic regression derivatives for w_j and b are different, when where is an x_j[i] outside of the brackets for w_j. So my question is why why do we have x_j[i] multiplication for w_j and do not have for b?

2

That’s how the result becomes when you do the calculus for the partial derivative of the cost with respect to b.

Intuitively, if you look at f_wb = w*x + b, you can see that ‘w’ is scaled by x, but ‘b’ is not. This is the basis for why dj_dw and dj_db have different forms.

1 Like
3

Thanks for your reply. Intuitively, I’ve also thought about the absence of x in the b part, however, after your reply I have confidence.