I have honestly spent several days reading pdfs over pdfs to try and understand this. I'm so glad I found this video!
@kniti27x
5 жыл бұрын
If anyone finds this too complex, I can recommend the Lecture 16 about SVM's from the AI at MIT (here on youtube)
@akshitagarwal98
5 жыл бұрын
Thank for the info.
@emmittjesse7443
3 жыл бұрын
I know Im asking the wrong place but does someone know a tool to get back into an instagram account?? I somehow lost my account password. I appreciate any help you can give me!
@dimitrab6485
6 жыл бұрын
Nice video, but a bit advanced, it presupposes someone already understands a lot about SVMs and optimization.
@antonschaefer7116
6 жыл бұрын
Thank you for the series on SVMs it's really helpful. At around 2:23 the slide says g_i alpha_i = 0 [...] Shouldn't it be g_i < 0 --> alpha_i = 0, g_i >= 0 --> a_i -> inf instead, so that a point right on the boundary is counted with an alpha_i > 0?
@calberk1737
Жыл бұрын
Why does a_i go to infinity in the first place?
@koizikang1594
4 жыл бұрын
Hi Prof. Thanks for the video. It helps a lot for my understanding. May I ask why is there a 1/2 at 3:29 in front of the f(theta)?
@Babykiyaan3jan2023
5 жыл бұрын
The video is really informative...understood how the SVM is working but i am trying to implement it in verilog which does not contain any in built library for the same...could you please help me with that.
@tuliomoreira7494
5 жыл бұрын
Great video, although I have the same question as Anton Schaefer (about non-zero alpha values for g_i > 0. At the margin we should have w*x + b = 1 (for y_i = 1). Then g_i should be 0. I don't get the g_i > 0 part though... from my understanding this would only happen at the decision boundary itself (w*x + b = 0), where we we would get g_i = 1
@dylanbarratt5717
3 жыл бұрын
Great point, Tulio!
@achyutanandasahoo4775
6 жыл бұрын
How to choose C parameter ( penalty parameter ) in multi class SVM classifier ? I know cross validation can be used to find optimal C value. But point is what should be my initial C value ? Like where to start ? Like 0.1 or 10 or 50 or 100 or 300 or 500. ..etc
@sniper9143
5 жыл бұрын
Hello for pardon my ignorance but how can i plug givens back into the dual form to find w and b?
@yasserothman4023
3 жыл бұрын
@7:22 why is epsilon subtracted from 1 ?
@ahmadmohamad4567
8 жыл бұрын
Hello Thank you very much Would you please give me an example for tow points like (v1(-1, -1) output(1) , v2 (1,1) output (1) ) and show me how can i apply the Lagrangian Dual Problem
@heribertoenlo
7 жыл бұрын
Hi, I'm trying to find the dual problem for an adaptive SVM . Please help. I'm leaving a link to the problem. stats.stackexchange.com/questions/270017/adaptive-svm-formulation-finding-the-dual-of-the-optimization-problem
Пікірлер: 17