This course aims to teach learners how to build versatile, data-efficient, and generalizable machines that can learn to see, reason about, and interact with the physical world. The course covers topics such as exploiting generic causal structures, integrating knowledge from computer graphics, physics, and language, and utilizing deep learning. The teaching method involves a seminar-style presentation by an expert in the field. The intended audience for this course includes individuals interested in computer science, artificial intelligence, machine learning, and physical scene understanding.
Overview
Syllabus
Introduction
Building infant machines
Scaling up data
Challenges
Inversion
Intermediate Representation
Physical Model
Recap
Learning to Augment
Generation of Average Level Networks
Augmented Graphics
Dynamic Engine
Physics Engine
Summary
Example
Key principle
Collaborators
Questions
Dynamics
Audience Questions
Object as Parts
Other Physical Properties
Taught by
Stanford HAI
