Skip to content

Rich Sutton: “Myths of Representation Learning” [video]

youtube.com
3 pointssinwave1 comment
On HN

Comments

Hilarious moment at 36:00. Sutton has just finished talking about how in real life, functions that learning agents wish to estimate will often be non-stationary because life is constantly changing.

An audience member complains that this is unfair (i.e. unfair to the learning algorithm), and somebody nearby the cameraman retorts "life is unfair".

This is a fact that even algorithms cannot avoid.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.