
[10] Chelsea Finn - Learning to Learn with Gradients
The Thesis Review
How to Learn a Classifier That Tells the Robot Whether It Succeeded at a Task
Roblox developed an algorithm that could learn a classifier to tell the robot whether it succeeded at a task. They collected both positive and negatives of different kinds of tasks like putting. It then did a kind of meta-learn algorithm such that when it ran gradient descent with respect to just a few positive examples of the task about five I think then it could learn aclassifier that told you success or failure for that task.
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.