Judea Pearl's new paper on the limitations of machine learning (namely that it can't reason about causality) is interesting. It feels like there ought to be an effective way to incorporate do-calculus into deep networks. https://t.co/EcrVzj3zqZ
— Bharath Ramsundar (@rbhar90) January 15, 2018
from Twitter https://twitter.com/rbhar90
January 14, 2018 at 08:52PM
via IFTTT
No comments:
Post a Comment