Information in Infinite Ensembles of Infinitely-Wide Neural Networks
Ravid Shwartz-Ziv, Alexander A. Alemi · arXiv (Cornell University) · 2019
In this preliminary work, we study the generalization properties of infinite ensembles of infinitely-wide neural networks. Amazingly, this model family admits tractable calculations for many information-theoretic quantities. We report analytical and empirical investigations in the search for signals that correlate with generalization.