History
1998 LeCun - digits
2012 AlexNet - Krizhevsky et al.
ZFNet
2013
2014 - more efficient
2015 Residual Networks vs Plain net architectures
you can do more dropout
2016 - scale down, go wide
ILSVRC
2016 - Transformers
https://arxiv.org/pdf/1706.03762.pdf