[10] Chelsea Finn - Learning to Learn with Gradients
The Thesis Review
00:00
The Importance of Tasks in Machine Learning
In imitation learning he typically you need to collect demonstrations which is like teleoperating a robot to do a certain task. So in this work kind of the way that we looked at is we wanted to see if a robot could quickly adapt to a new task and learn to perform a new task with just a single demonstration. The motivation again is that if you want to train a robot usually you to be trained from scratch it requires a fair amount of data. It would be something like using learning how to like kind of use a specific object in a specific context.
Transcript
Play full episode
