[10] Chelsea Finn - Learning to Learn with Gradients
The Thesis Review
00:00
How to Learn a Reward Function With Inverse Reinforcement Learning
Med learning gives you a mechanism to encode prior information. If you have a prior over what the human is trying to do and what sort of tasks are interesting and meaningful then you should be able to much more effectively in further intent so that's kind of an area where I think that actually med learning or some form of learning priors makes a lot of sense.
Transcript
Play full episode
