Bayesian Deep Learning
Avik Santra, Souvik Hazra, Lorenzo Servadei, Thomas Stadelmayer, Michael Stephan, Anand Dubey · 2022
This chapter will give an overview on the fundamental principle of learning theory. Later, keeping goal of learning theory, formulation of deterministic architecture derived followed by its inherent limitation on representation learning. These limitations are due to model architecture and imitated or noisy measurement data. This leads to formulation of uncertainty with in deterministic neural networks and gives path to Bayesian neural network. This chapter provides detailed understanding on Bayesian learning followed by understanding on different elemental blocks required to formulate Bayesian deep learning. Later, different optimizing techniques are compared to realize Bayesian deep learning. At the end, a practical application is demonstrated on Bayesian deep learning for an automotive radar using state-of-the-art optimization technique.