Minicourse: Regularization Methods for Machine Learning

Minicourse: Regularization Methods for Machine Learning
Speaker:  Sergei Pereverzyev - RICAM, JK University
  Monday, April 23, 2018 at 8:30 AM
In this course we offer a well-balanced mixture of basic and innovative aspects in regularisation methods for inverse and ill-posed problems.
We will demonstrate new, differentiated viewpoints, and important examples for applications. Some of the basic and current developments
in the field of regularization theory will be explored, such as multiparameter regularization and regularization in learning theory.


The course will mainly follow the Chapters 2 and 4 of the book "Regularization Theory for Ill-posed Problems: Selected Topics",
and will be divided in 8 lectures as follows:
  1. 23/04,  "Mathematical Aspects of Data Science. Supervised learning as an ill-posed inverse problem."
  2. 24/04, "Basics of the Regularization theory. Single-parameter regularization schemes"
  3. 26/04, "Kernel Ridge Regression and beyond. Learning rates for regression and ranking settings".
  4. 27/04, "Data-driven choice of regularization parameters. Balancing principle & Co".
  5. 30/04, "Aggregation of regularized learners. Linear functional strategy"
  6. 02/05, "Multiple kernel learning. Illustration by an application to diabetes technology"
  7. 03/05, "Multiple penalty regularization and semi-supervised learning"
  8. 04/05, "Some topics that were left out of the course"
Instructor: Prof. Dr. Sergei Pereverzyev
Ricam, Johann Radon Institute, Linz, Austria 


For a detailed  time table:  (TBA).



 

Programme Director
Giacomo Albi

External reference
Publication date
March 1, 2018

Studying

Share