Understanding_Deep_Learning_Requires_Rethinking_Generalization.md

Understanding Deep Learning Requires Rethinking Generalization

TL;DR: Deep neural networks easily fit random labels, and we need better ways to analyze generalization error.

See this blog post I wrote. These notes will not be updated on GitHub.