5:30 PM - 6:10 PM
[2Q6-OS-20b-01] (OS invited talk) Principle-centric AI for Embodied Intelligence
Keywords:Deep learning, Computer vision
How can we get Deep Learning's success on the web to translate to intelligent machines physically interacting with the real-world, especially for safety-critical systems like cars and robots? In this talk, I will describe our unique approach to the problem of embodied intelligence: Principle-centric AI. It consists in going beyond mere data, the main workhorse of DL, by complementing it with guiding principles to increase safety, sample efficiency, and robustness. I will showcase a few examples of cutting edge computer vision research showing how to leverage geometric principles as self-supervised objectives and inductive priors to not just let machines learn by examples, but actually guide them as teachers towards better outcomes.
Authentication for paper PDF access
A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.