Kernels Introduction - Practical Machine Learning Tutorial with Python p.29

In this machine learning tutorial, we introduce the concept of Kernels. Kernels can be used with the Support Vector Machine in order to take a new perspective and hopefully allow us to translate into further dimensions in order to find a linearly separable case.

[shauryavardhan]

2 Million watched the first video of this series.
Congrats! you are among the top 80k survivors.

[classicrockman]

The use of dot products in this video is a bit overused. Your alphas and y's are NOT vectors, therefore you can't dot them with anything. Your x's and w's are the vectors. In your Lagrange equation only the x's at the end are dotted. Those are the N features in N dimensions for both the positive and negative classes. The alpha's are just a weighting and the y's are just a binary classification.

The reason SVM's are powerful is due to only relying on a dot product in the Lagrangian and therefore you can transform the x's to higher dimensions where classification is easier but still yield a scalar value from the dot product for optimization.

Watching the MIT video on SVM's clarified this more for me.

[binus]

I first came in contact with kernels while reading the electrostatics in the high school.

[binus]

what to do when the realities of the world hits u???we use kernal boiss...thanks sentdex for the wise advice.

[abbasnosrat]

For image classification you could use pca modes with lower energy that contain more detail and separate better

[gokan]

Divinely detailed explanation!! Keep posting, enjoyed much!
Thanks @sentdex! ;)

[Steven-dc]

Honestly loving the IBM Watson Ads I see on your channel

[bejgaonnischit]

@11:30 & @12:24 it is not dot product don't get confused. @12:30 it is dot product

[algar]

Your videos are great and helped me so much understand svm and learn how to use it, with detailed explanations. You are the best. One thing that I am still struggling to do is to use SVM/Kernels for regression (not just classification) on strings (such as peptides). I have not found anything close to that in your videos, and everything else out there is too obscure for me to learn to actually use. Could you direct me to any source you have found that explains it simply enough or even better do it yourself ! Thanks again so much. Alain

[nikolahuang]

I clicked thumb right after I listened to the sound of Bucky Robert. It turns out that Bucky just won't disappoint us.

[sukritgupta]

when the reality of the world hits you... damn straight

[ammmyu]

Hi sentdex, I am working on a project to detect cancer in lung ct scan images, I dont know how to extract features from the image using python, so that i can feed that to my classifier. I have already done preprocessing steps of image conversion to binary and used ostu's method. Can you help me

[Maisonier]

Is It Possible to build a kernel with AI ?

[yashumahajan]

please add some tutorials on Tenforflow from basics to advance. just like SVM

[chandeepadissanayake]

Multiplication between Alpha sub i and Y sub i is not a dot product mate... Oh the Commentary over the Math!

[MrKirakh]

Hi Sentdex, the whole series is very helpful and easy to follow. Thanks a lot.
I was wondering if you have such a video on Kernel ridge regression? Are you planning to any soon?

[shashikantkharwar]

Good content. Keep going. We'll support you.

[luojihencha]

Before blablab shouldn't you introduce the definition first?

[stanlukash]

This video made me even more confused.

[focker]

I am surprised by the fact that the nb of views of this video is so low. Kernels is such an important thing in svm