What is a kernel in a Support Vector Machine algorithm?




I am not able to understand the concept of kernels in the Support Vector Machine(SVM) algorithm. What exactly is a kernel here and what is its use and function?



@kunal @aayushmnit @ajay_ohri @tavish_srivastava, guys if you can share your knowledge on this particular area that would be a great help. I’m also facing similar understanding dilemma.



A kernel is simply a means of projecting into a higher dimension. Check this brilliant video and it will be hopefully clear. https://www.youtube.com/watch?v=3liCbRZPrZA


Hi Aditya, devansh,

Technically : Kernels are internal functions which are tied up by the learning coefficients to form a cost equation which you optimize. The most popular kernel types(in descending order) are -Gaussian, Linear, Polynomial Kernels etc.

Lehman Language : These are small-small functions which your model create and throw on your dataset to fit the data accordingly (Just like a K means clustering works but in a supervised way). These functions have shaped like gaussian , linear line, or polynomial function.

For more appreciation over the topic I am giving link to Machine learning class on coursera taken by Andrew NG, module 12 (SVM) - Lecture 4 & 5 (Kernels - I & II) -
Lec 4 : https://www.youtube.com/watch?v=5yRSFkWAe3E
Lec 5 : https://www.youtube.com/watch?v=mKT5CSBhXfU

Hope this helps.

Aayush Agrawal