Anca Dragan (UC Berkeley): "An Optimization-Centric Theory of Mind for Human-Robot Interaction" - 2019

Title : Anca Dragan (UC Berkeley): "An Optimization-Centric Theory of Mind for Human-Robot Interaction" Author(s): MIT Institute for Data, Systems, and Society Link(s) : https://www.youtube.com/watch?v=AfrrswdSHM4

When an AI agent is involved in an environment with a human also interacts with the environment, we will have a utility function that depends on not only the AI agent's actions, but also on the human's actions. The AI agent thus needs to reason about what the human's policy is in this Multi-Agent Reinforcement Learning setting.