392195 Vertiefung Neuronale Netze (V) (WiSe 2025/2026)

This lecture will consider various unsupervised and instance-based learning approaches before turning back to deep neural networks. In the exercises will replicate various classical applications of neural network methods to interesting real-world problems.

• Topological Maps: Self-Organizing Maps, Growing Neural Gas, Hyperbolic SOM, Local Linear Maps
• Mixture Models: Gaussian Mixture Models, Gaussian Mixture Regression, Gaussian Processes
• Recurrent Neural Nets: Dynamics, Stability, applications: Hopfield net, CLM, ...
• Graph Networks: Processing Structural Information
• Generative Models: Generative Adversarial Networks (GAN), Actor-Critic, Variational Autoencoders
• Deep Reinforcement Learning

The lecture will be given in English language if desired by the audience.
Slides will be in English in any case.

Requirements for participation, required level

recommended prerequisites:
- Machine Learning basics
- Neural Networks basics

External comments page

http://ni.www.techfak.uni-bielefeld.de/teaching/vertiefung-neuronale-netze