Regularization of Kernel Methods by Decreasing the Bandwidth of the Gaussian Kernel

Posted in Science on August 03, 2008


Regularization of Kernel Methods by Decreasing the Bandwidth of the Gaussian Kernel

Lecture slides:

  • Outline
  • Gaussian kernel and RKHS
  • Gaussian RKHS
  • Learning in Gaussian RKHS
  • Motivation 1: The effect of regularization
  • Motivation 2: One-class SVM
  • Outline
  • Setting and notations
  • Intuitive behavior
  • Main result: consistency
  • Main result: asymptotic shape
  • Application: two-class SVM
  • Application: one-class SVM
  • Conclusion

Author: Jean-Philippe Vert, Ecole des Mines de Paris - Paris Tech

Watch Video

Tags: Science, Lectures, Computer Science, Machine Learning, VideoLectures.Net, Gaussian Processes, Kernel Methods