inControl cover image

ep3 - Ben Recht: A tour of optimization, machine learning, and control

inControl

00:00

Optimism in the Face of Uncertainty

When you have very little data, you did need to have some kind of quantification of the uncertainty in order to get stable performance. But as you start to just get a reasonable amount of data, what you could just do is fit the model using least squares and then plug the model in as if it was true. It's surprisingly powerful. And I believe in model predictive control. This is called disturbance feedback. There's definitely something where you operate with this map that maps the error disturbance into your state and theerror disturbance into your controller.

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