122 - Building Conversational AI's with Joe Bradley
Programming Throwdown
00:00
How to Handle Uncertainty in Deep Networks
I think as a field, handling uncertaintyis is very, very difficult right now with deep networks. I mean, there's effectively two or three amps. There's one camp that says, well, train ten different models. And then there's this camp that says,. let's put priors on everything and use conobasian approachesad, none of them work very well. The reason why you can't just go on pytorch dot org or tenser flow dot com, or whateverd is because s not, ah, really a method that satisfies it very well.
Transcript
Play full episode
