Probabilistic Deep Learning with TensorFlow 2

Probabilistic modelling is a powerful and principled approach that provides a framework in which to take account of uncertainty in the data.

This module focuses on using probabilistic layers from TensorFlow Probability to develop deep learning models that quantify uncertainty in both data and model predictions, particularly important for safety-critical applications.

Normalising flows are covered as a powerful class of generative models that model data distribution by transforming simple base distributions through bijective transformations.

This module covers implementing variational autoencoders using TensorFlow Probability, including encoder/inference networks and decoder/generative networks. Students learn to encode data into compressed latent space and generate new samples.

The Capstone Project brings together concepts from the course including Distribution objects, probabilistic layers, bijectors, and KL divergence optimization.