theory

Neural Weight Initialization and Bayesian Priors: Unfrightful Story of a Doppelgänger

Weight initialisation is key to both semi-supervised learning and meta-learning. What is its relationship to Bayesian priors?