Tweeted By @JeffDean

Based on this research work:

AutoAugment: Learning Augmentation Policies from Data
Ekin D. Cubuk, Barret Zoph, Dandelion Mane, Vijay Vasudevan, Quoc Lehttps://t.co/O3rZLcECzg

— Jeff Dean (@JeffDean) October 27, 2018