Gaussian Processes for Machine Learning [2018 SoSe] | ||
---|---|---|
Code IGPML |
Name Gaussian Processes for Machine Learning |
|
LP 5 LP |
Dauer one semester |
Angebotsturnus irregular |
Format Lecture 2 SWS |
Arbeitsaufwand 150 h; thereof 30 h lecture 100 h project 20 h report |
Verwendbarkeit B.Sc. Angewandte Informatik, M.Sc. Angewandte Informatik, M.Sc. Scientific Computing |
Sprache |
Lehrende |
Prüfungsschema |
Lernziele | To build a solid background on both the theory of Gaussian processes (GPs) and how they are used in practice to build effective machine learning models. * Firm theoretical knowledge on how to use GPs for machine learning. * Knowledge on how big data can be modeled with GPs. * Practice on how to design, develop, and evaluate a powerful machine learning model. |
|
Lerninhalte | This module covers the following topics: * Introduction to and motivation for GPs. * Predicting real-valued and categorical output with GPs. * Approximate inference of GPs. * Modeling big data with GPs. * Exploratory data analysis and knowledge discovery with GPs. * Time series modeling with GPs. * Deep learning with GPs. * GPs for alternative learning setups. |
|
Teilnahme- voraus- setzungen |
recommended are: basic background on probability and statistics, basic knowledge on machine learning and linear algebra | |
Vergabe der LP und Modulendnote | Bestehen der Modulprüfung | |
Nützliche Literatur | Carl E. Rasmussen, Christopher I. Williams, Gaussian Processes for Machine Learning, MIT Press, 2006 (online) Christopher M. Bishop, Pattern Recognition and Machine Learning, Springer, 2007 |