Lex Fridman Podcast cover image

Juergen Schmidhuber: Godel Machines, Meta-Learning, and LSTMs

Lex Fridman Podcast

00:00

Is It Necessary to Look Back in Time?

Eliteum is one of the great examples of architectures that a do something beyond just deeper and deeper networks. It's not clear what are the practical limits of the elustien when it comes to looking back. The credit assignment problem across way back, not just 50 time steps, or a hundred or a thousand, but millions and billions. So essentially, although it's predicting only the next time step, it is motivated to learn,. to put into memory something that happened maybe a million steps ago - because it's important to memorize that if you want to predict that at the next timestep, the next evento.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app