Hacker Read

mjfl · 2019-02-03 02:31:21+00:00

isn't it just gradient descent?

jofer | karma 3770 | avg karma 6.5 · | 2016-08-05 15:05:56

Gradient descent is for non-linear problems where you can't directly invert or somehow linearize the problem.

bra-ket | karma 2953 | avg karma 2.75 · | 2017-09-01 03:43:19

does it mean we don't need gradient descent after all to achieve the same result?

GFK_of_xmaspast | karma 3102 | avg karma 0.89 · | 2016-05-25 14:20:02+00:00

I know what gradient descent is, thanks, I was referring to the rest of that mess.

jdrc | karma 1088 | avg karma 1.79 · | 2022-04-23 18:03:47

"gradient descent" should be enough

UncleOxidant | karma 8563 | avg karma 3.16 · | 2022-01-07 12:08:32

Gradient descent in action.

TeMPOraL | karma 106045 | avg karma 3.04 · | 2023-10-26 05:41:29

Or gradient descent if you mentally negate the number in question. It's the same thing.

Iv | karma 9755 | avg karma 4.1 · | 2020-12-06 16:02:00+00:00

Thing is, gradient descent is not really a complex algorithm.

anigbrowl | karma 91010 | avg karma 3.36 · | 2023-02-16 16:01:07

No. I think you should rework your gradient descent algorithm, it's bad.

jph00 | karma 7468 | avg karma 8.69 · | 2017-07-23 04:41:12+00:00

Gradient descent is covered. We even show a pytorch implementation :)

darpa_escapee | karma 1683 | avg karma 1.96 · | 2019-07-13 22:51:33

AI is gradient descent?

sjtrny | karma 585 | avg karma 1.68 · | 2014-06-27 01:32:35+00:00

Accelerated gradient descent is pretty common everywhere.

clairity | karma 9767 | avg karma 2.26 · | 2020-02-12 18:29:21

note that "gradient descent" isn't AI either. it's more computational linear algebra: a heuristic for numerical methods used to solve (usually to a local extrema) systems of equations without direct analytical solution(s).

kragen | karma 31428 | avg karma 2.09 · | 2022-03-30 16:55:25

People mostly use gradient descent to "solve" nonconvex problems.

jtmcmc | karma 571 | avg karma 1.75 · | 2016-05-21 18:11:01+00:00

They're all various optimization technique, variations on gradient descent.

deepnotderp | karma 4485 | avg karma 2.3 · | 2019-11-15 06:44:57+00:00

Yeah, it's quite similar to "Learning to learn by gradient descent by gradient descent" and related works

rsrsrs86 | karma 149 | avg karma 0.84 · | 2017-06-29 18:38:09

What he calls back propagation is actually gradient descent.

mav3r1ck | karma 301 | avg karma 5.79 · | 2018-03-22 16:58:21+00:00

You are not off base at all, thanks for clarify and sorry for the confusion, I did not mean to say it was using gradient descent. It's been a while. The term I was thinking of was multiple "simulated annealing".

YeGoblynQueenne | karma 22041 | avg karma 2.5 · | 2018-03-02 11:23:09+00:00

Most likely, gradient descent with momentum.

modeless | karma 36822 | avg karma 6.69 · | 2021-12-10 18:53:30

So it doesn't use a neural network, but it is still optimized by gradient descent. Differentiability is the key!