These are solutions for a course at the TU Berlin 2019/2020.
The topics of the sheets are as follows:
- Visual Perception
- Slow Feature Analysis
- Learning Dynamics in Deep Linear Networks
- Saliency Maps
- Perceptual Bistability
- Decision Making: Extension-Drift Model
- Decision Making: Chapman-Kolmogorov
- Reinforcement: Multi-Armed Bandit
For all results, see the respective jupyter notebooks. See, for instance:
2. Slow feature analysis
4. Saliency maps
The jupyter notebooks are not dependent on each other and can be run individually. Most of them need NumPy and Matplotlib, so you might want to install these. Rarely, SciPy and PyTorch are also used.