Momentum in SGD|Understanding Momentum in stochastic gradient descent

[@honprarules]

Clearest mathematical description on the internet. Trust me, I have searched a LOT and I am glad I found this hidden gem!

[@UnfoldDataScience]

Glad it was helpful Tarun :)

[@rahuldey6369]

Brother I've come here to clear the concept on SGD with momentum after watching 3-4 videos in some well known channels,but this is so far the best explanation I've got. Thank you. Great work. Subscribed

[@UnfoldDataScience]

It's my pleasure Rahul. Thanks a lot.

[@rahuldey6369]

@@UnfoldDataScience Just a small suggestion,you may try to complete the playlist on deep learning,so that is a flow people are able to get the concepts in an order. I can understand how much time does it take to cover a topic and make a video, but it'll be helpful. I really appreciate your work

[@tintuayan]

As beta is 0.9, it is giving more importance to the the older values and less importance to the newer values but you have said the opposite.

[@UnfoldDataScience]

Is it, let me check. Thanks a lot.

[@avinashajmera2775]

@@UnfoldDataScienceplease give clarification

[@rishigupta2342]

Very good explanation!!!. So sgd contains mini batch and momentum.

[@bangarrajumuppidu8354]

great explanation with deep understanding of momentum

[@UnfoldDataScience]

Thanks alot

[@sandipansarkar9211]

great explanation

[@UnfoldDataScience]

Thanks again.

[@AlgoTribes]

your explanations are as addictive as watching Netflix series..you can't leave for tomorrow..😀😀

[@UnfoldDataScience]

Glad you like them Saurabh. Your comments are my motivation :)

[@mpthorat]

nice explanation

[@RupeshKumar-bu2xy]

In the last equation is V the smoothig factor and dw is common in all so will it get cancelled or this equation is Vdw(t)= BVdw(t-1) + (1-B)dw(t)

[@shanmukhchandrayama8508]

can someone answer this

[@ravindrav1895]

I loved ur video sir , I understood everything in this video , thank you sir

[@UnfoldDataScience]

Thanks for your feedback. Please share with others as well.

[@sowdanyasekar9854]

Really helpful..pretty good explanation..👍

[@UnfoldDataScience]

Glad it was helpful Sowdanya :)

[@phanikumar3136]

Clear cut explanation sir.... 👌

[@UnfoldDataScience]

Thanks Phani. Happy Learning. tc

[@MrFringon]

Muy claro, muchas gracias

[@UnfoldDataScience]

Eres bienvenido

[@adityasharma2667]

Hi Aman, can you please make video on Transformers and BERT

[@김수연-r8c]

really great explanations !!

[@UnfoldDataScience]

Glad you think so!

[@lanciauxrayzimba]

thanks for the info. what will happen if you put beta = 1?

[@UnfoldDataScience]

More aggressive steps in learning.

[@akansh_26]

Please make a video on Logging meaning and basics in python

[@UnfoldDataScience]

Hi Akansh, thank you for your ask. Noted.

[@pnachtwey]

The example is not reality. Real data in not like minimizing f(x,y) = x^2+y^2 or even the Rosenbrock formula. In reality the terrain is more like the Grand Canyon where the correct path is very narrow and any direction will often hit a canyon wall. I also object to instructor that don't have working examples. Writing on a whiteboard doesn't hack it anymore.