[24] Martin Arjovsky - Out of Distribution Generalization in Machine Learning
The Thesis Review
00:00
Optimization and the Out of Distribution Generalization
were 30 different problems that appeared from the node that had to do with okay how do we optimize this what does the optimization landscape look like. It's always both I guess I'm sorry for the lame answer but you know it's always a bit of both, he says. "I think something that I've done throughout is always kind of keep the problem first and make networks generalize better in practice"
Transcript
Play full episode
