Assignments for the Probabilistic Programming class taught at the Artificial Intelligence master program at University of Bucharest.
All notebooks are opened in Google Colab.
PyMC 2 implementation of the LDA algorithm for topic modeling.
Notebook: Latent Dirichlet Allocation - PyMC
Implemented multiple bayesian neural network variants in Pyro, a probabilistic programming framework with PyTorch backend.
Notebooks bellow:
Bayesian Fully Connected Neural Networks:
- Bayesian Neural Network - Sanity Check
- Bayesian Neural Network - Binary Classification (Moons)
- Bayesian Neural Network - Multiclass Classification (Blobs)
Bayesian Convolutional Neural Network:
Bayesian Siamese Neural Network (based on the method described by Koch et al in Siamese Neural Networks for One-shot Image Recognition):