Kernel Methods

Introduction to Machine Learning - 10-701/15-781

Content

  • Kernel properties

    • Hilbert spaces

    • Convex cone, distances, norms

    • Mercer's theorem

    • Representer theorem

    • Kernel expansion

    • Kernel trick in general

  • Regularization and features

    • Shotgun approach

    • Regularization

    • Explicit feature map

  • Hash kernels

    • Application - spam filtering

    • Hashing trick

  • Application - jet engine failure detection

  • Support Vector Novelty Detection

    • Density estimation

    • Thresholding and scaling

    • Optimization problem and dual

    • Online setting

  • Support Vector Regression

    • Loss functions

    • Optimization problem and dual

    • Examples

  • Kernel PCA

    • Principal Component Analysis

    • Kernelization

Supplementary material

Slides in PDF and Keynote. If you want to extract the equations from the slides you can do so by using LaTeXit, simply by dragging the equation images into it.

Videos

This is unedited video straight from a Lumix GF2. It should help as a supplement with the slides.