[24] Martin Arjovsky - Out of Distribution Generalization in Machine Learning
The Thesis Review
00:00
Is There a Gap Between IRM and IRM?
Do you think that there's some gap between what you've shown theoretically and what happens in practice or is it kind of yeah like how do you think about thatYeah so it's a good question i'm gonna be very honest i remember is useful only in very few instances of like the spoolish correlation problems that we were wanted to solve. There have been industrial applications and stuff nice eyes using it for for stuff related to discovery of cancer uh in astronauts and stuff but most of the times that people try to use it it doesn't really work.
Transcript
Play full episode
