Abstract
In the previous chapters, we described the Support Vector Machines as a method that creates an optimal hyperplane separating two classes by minimizing the loss via margin maximization. This maximization led to a dual optimization problem resulting in a Lagrangian function which is quadratic and requires simple inequality constraints. The support vectors are responsible for defining the hyperplane, resulting in the support vector classifier f which not only provides a unique solution to the problem, but also ensures learning with tighter guarantees. However this is only possible if the input space is sufficiently linearly separable. On the other hand, many input spaces are, in fact, not linearly separable. In order to overcome this restriction, nonlinear transformations can be used to implicitly obtain a more adequate space.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
There are variations for PCA to study data subspaces which employ XX T in order to analyze the correlations among position vectors.
- 2.
Recall a Hilbert space is simply a vector space for which: a distance metric, a vector norm and a dot product are defined. Because of that, it supports projection of vectors into spaces, and therefore geometry.
- 3.
Please, refer to the MNIST database http://yann.lecun.com/exdb/mnist.
References
R. Courant, D. Hilbert, Methods of Mathematical Physics. Methods of Mathematical Physics, vol. 2 (Interscience Publishers, Geneva, 1962)
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG, part of Springer Nature
About this chapter
Cite this chapter
Fernandes de Mello, R., Antonelli Ponti, M. (2018). A Brief Introduction on Kernels. In: Machine Learning. Springer, Cham. https://doi.org/10.1007/978-3-319-94989-5_6
Download citation
DOI: https://doi.org/10.1007/978-3-319-94989-5_6
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-94988-8
Online ISBN: 978-3-319-94989-5
eBook Packages: Computer ScienceComputer Science (R0)