Information in Infinite Ensembles of Infinitely-Wide Neural Networks
2019-11-20pproximateinference AABI Symposium 2019Code Available0· sign in to hype
Ravid Shwartz-Ziv, Alexander A. Alemi
Code Available — Be the first to reproduce this paper.
ReproduceCode
Abstract
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.