A Kernel-based Support Vector Machine by Peter Axelberg and Johan Löfhede
Maximum Margin Classifier Decision boundary 22 11 Maximum Margin
Need for more complex decision boundaries In general, real-world applications require more complex decision boundaries than linear functions.
Dimension expansion The SVM offers a method where the input space is mapped by a non-linear function,, to a higher dimensional feature space where the classes are more likely to be linearly separable. (x)(x)
Input space High dimensional feature space x1x1 (x1)(x1) (Xi)(Xi) Decision space 11 22 xixi (Xi)(Xi) f( ( X i )) High dimensional feature space
Kernel function Instead of directly using the mapping vectors in the high dimentional feature space, a kernel function K(x,x i ) can be introduced in the input space according to the following substitution: (x)(x) where denotes the inner product
The Kernel function used by the decision boundary function
Examples of Kernel functions Polynomial Radial basis function (RBF) Sigmoidal
A real world example Classification of some fruits/vegetables using kernel-based SVM
Feature 1Feature 2Feature 3Feature 4
Circumference 1 (longest) (cm) Circumference 2 (shortest) (cm) Weight (g)Color (code)