Tweeted By @dustinvtran
One year later with revisions, and it's at UAI this year. For those interested, we included a formal derivation of how training neural nets w/ NCP corresponds to variational inference with a KL in output space instead of weight space (Sec 3). https://t.co/2YKu0WVOt3 https://t.co/JIeteYVluf
— Dustin Tran (@dustinvtran) July 2, 2019