How would 1D gradient descent look like?

Question

We have always known that gradient descent is a function of two or more variables. But how can we geometrically represent gradient descent if it is a function of only one variable?

score 2 · Answer 1 · answered Mar 30 '20 at 21:51

For a function of one variable, there are only two options for directions in the domain: left or right, so it becomes almost trivial, but you can still talk about gradient descent.

You would take steps to the left if the slope/derivative is positive and make steps to the right if the slope/derivative is negative--i.e. the opposite direction of the derivative (the 1d version of the "gradient" in gradient descent), which is equivalent to the higher dimensional case.

score 0 · Answer 2 · answered Jul 18 '17 at 14:36

0

"The concept of a direction of fastest descent only makes sense in more than one dimension."

https://math.stackexchange.com/a/180573

answered Jul 18 '17 at 14:36

David Duffrin

17
1

I think this answer is not correct. See the other answer: https://ai.stackexchange.com/a/19883/2444. – nbro Mar 30 '20 at 23:25

How would 1D gradient descent look like?

2 Answers2