[10] Chelsea Finn - Learning to Learn with Gradients
The Thesis Review
The Evolution of Neural Networks in Meta Learning
At the time I guess like you mentioned some of the some of the previous work. But I guess like this idea of doing it with gradients was maybe the new thing. So how did you kind of like rule out did you try out the existing approaches or did you know that you needed something kind of new? Yeah so the existing approaches weren't either primarily either using kind of evolutionary methods a lot of them were not really demonstrated on on large scale problems they're kind of demonstrated on very close problems. And so the key idea behind the model of NOSIC meta learning algorithm which is the first work that we did in like late 2016 early 2017 was to try to embed gradient descent inside the
00:00
Transcript
Play full episode
Remember Everything You Learn from Podcasts
Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
