What stops gradient descent from finding the largest error?

Question

If a gradient points towards a max or a min what stops gradient descent from maximizing error instead of minimizing it?

Is it the nature of the update step that makes this process one way?

Possible duplicate of How can change in cost function be positive? see also https://stats.stackexchange.com/questions/367397/for-convex-problems-does-gradient-in-stochastic-gradient-descent-sgd-always-p/367459#367459 — Sycorax, Apr 22 '19 at 17:19

score 0 · Answer 1 · answered Apr 22 '19 at 16:58

0

In gradient decent, we are following the negative gradient direction, where the objective function will decrease instead of increase.

answered Apr 22 '19 at 16:58

Haitao Du

1 Answers1