
Episode 15: Martín Arjovsky, INRIA, on benchmarks for robustness and geometric information theory
Generally Intelligent
00:00
The Uncertainty in Prediction
When the classifie is uncertain, but all suggest and supervise, you have a bunch of lak points. You want to say, when something looks queervac analamous human can take a look at it. I think this is very underrate for good reasons. It's hard to do research when you cannot convince any onethat what your saying is working. And i don't have any good ideas how to solve this either.
Transcript
Play full episode