SLIDE 1
Support Vector Machines & Kernelization Barna Saha Most of the - - PowerPoint PPT Presentation
Support Vector Machines & Kernelization Barna Saha Most of the - - PowerPoint PPT Presentation
Support Vector Machines & Kernelization Barna Saha Most of the slides are made using David Sontags course on machine Learning at MIT Support Vector Machines Support Vector Machines What if the data is not linearly separable?
SLIDE 2
SLIDE 3
Support Vector Machines
SLIDE 4
Support Vector Machines
SLIDE 5
What if the data is not linearly separable?
5
- General idea: the original feature space can always be mapped to
a different (oDen some higher-dimensional feature space) where the training set is separable: [x1, x2]à[\sqrt(x1
2+x2 2), arctan(x2/x1)]
Φ: x → φ(x)
SLIDE 6
What if the data is not linearly separable?
6
- If there is a separator which “almost” separates, find a separator
that minimizes some kind of loss funcWon.
SLIDE 7
SLIDE 8
w is normal to the hyperplane w.x+b=0
SLIDE 9
SLIDE 10
SLIDE 11
SLIDE 12
SLIDE 13
SLIDE 14
SLIDE 15
SLIDE 16
SLIDE 17
SLIDE 18
SLIDE 19
ϕ(x1,x2)à(x1
2, x1x2, x2x1, x2 2)
SLIDE 20
SLIDE 21
Kernel method enables one to operate in a high-dimensional, implicit feature space without ever computing the coordinates of the data in that space but rather by simply computing the inner products between the images of all pairs of data in the feature space. Often computationally cheaper than the explicit computation of the coordinates.
SLIDE 22
SLIDE 23