Question regarding squared error cost and shape of J(w)

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

Hi

I noticed that, squared error function will always lead to J(w), which won’t have any other local minima, but only global minimum, why is that so ?
and, why for other cost functions (like others have been used in neural nets - no squared error), having multiple minima ?

Thanks

2

hi,
As per my knowledge, MSE is a type of quadratic equation. Quadratic equations always give only convex function(parabolic) so we have only one global minima.