WS 2007/2008

Neural Networks and Machine Learning

Lecture and Tutorial
Prof. Dr. Laurenz Wiskott

Tutorial: Thursday 14-16 o'clock in the 'Beratungsraum' of the Biology department, Invalidenstraße 43.
Lecture: Thursday 16-18 o'clock in the 'Beratungsraum' of the Biology department, Invalidenstraße 43.

This lecture covers methods of supervised and unsupervised learning and is intended to be a technically oriented complement to the Computational Neuroscience lectures. On the one hand we will learn about neural networks (e.g. backpropagation of error, reinforcement learning), on the other hand more abstract methods from Machine Learning will be presented (e.g. support vector machines, graphical models).

This course requires solid mathematical background in calculus and linear algebra. Some knowledge in probability theory is advantageous.

The default language for this lecture is English.

Lecture

1.	18.10.2007	Feedforward Neural Networks, Error Backpropagation
2.	25.10.2007	Generalization, Cross-Validation, Bias-Variance-Dilemma (Henning Sprekeler)
3.	01.11.2007	Support Vector Machines (Henning Sprekeler)
4.	08.11.2007	Nonlinear Expansion, Kernel Trick,
5.	15.11.2007	Bayesian Inference
6.	22.11.2007	Bayesian Learning
7.	29.11.2007	Inference in Bayesian Networks
8.	06.12.2007	Inference in Gibbsian Networks
9.	13.12.2007	Learning in Gibbsian Networks
10.	20.12.2007	Reinforcement Learning
11.	17.01.2008	Principal Component Analysis
12.	24.01.2008	Principal Component Analysis
13.	31.01.2008	Independent Component Analysis
14.	07.02.2008	Independent Component Analysis
15.	14.02.2008	Vector Quantization

Laurenz Wiskott, http://www.neuroinformatik.ruhr-uni-bochum.de/PEOPLE/wiskott/