Gradient Descent Part 2 Chieh

[@b.f.skinner4383]

This is one of the few videos on KZbin which really provides an in depth, formulaic tutorial on GD instead of a conceptual one. Thank you!

[@NiftyShifty1]

Great video. The matrix stuff on the side is really nice to see. Thank you

[@dareeul]

I have confuse about this topic so long ,but now you make me understand it.Thank

[@JohnSmith-lf5xm]

Hello Thank you for the explanation you are the only one that have explain it clear so far on 15 videos i have been watching. Thank you for real. I have a question.. I have the Gauss-Newton, Gradient Decent, and the Levenberg-Marquardt methods but I can not really understand the difference between them... can you explain that to me or tell me where to find it if you have other videos.

[@narasa12]

Excellent video... Thanks for posting

[@jamesspectre8492]

An excellent presentation! Brilliant!

[@吴志高-h9g]

thanks it makes me clear about gradient descent

[@43SunSon]

Question: where is the equation X_(i+1) = X_i-af'(X_i) coming from? Any resource I could take a look ? I wish to understand why the new one (X_i+1) is updated like that equation. Thank you!

[@Tothefutureand]

They call it math cause you can never know where it comes from

[@lindazuo9130]

Hi, Chieh, thanks for the video. Very clear. Just have one question: what is the step? Is the step along the x axis, y axis or the function curve? When you multiply the step and the first derivative of f(x), what do we get? I thought here you are trying to work out the gap between x1 and x2, but mathematically, it does not make sense to me. Many thanks if you could explain it.

[@muhammadseyab9032]

where is 1/2 in the derivative function?

[@rebeenali917]

Excellent

[@youssefdirani]

Thanks

[@ft1598128]

good job