Social Reinforcement Learning by Dr. Natasha Jaques

Date & Time: Wednesday, June 16th, 7 p.m. – 8 p.m. via zoom

Registration: https://uregina-ca.zoom.us/meeting/register/tJMsduGgqj4oHtNuAoc6wuDi8V0z44cPdCs0

Title: Social Reinforcement Learning

Abstract:
Social learning helps humans and animals rapidly adapt to new circumstances, and drives the emergence of complex learned behaviours. In this talk I will explore whether we can use insights from social learning to improve artificial intelligence (AI) algorithms. I will introduce Social Reinforcement Learning, which focuses on leveraging social learning to enhance AI learning and generalization, coordination with other AI agents, and human-AI interaction. For example, I propose new algorithms for enabling AI agents to learn from experts that are present in their environment, or learn how to have better conversations with humans by learning from their social cues. I also show how competition between agents can lead them to learn more complex behaviours, and generalize better to new tasks at test time. Overall, I argue that Social Reinforcement Learning is a valuable approach for developing more general, sophisticated, and human-compatible AI.

Background:
Natasha Jaques holds a joint position as a Research Scientist at Google Brain and post-doc at UC Berkeley. Her research focuses on social reinforcement learning -- developing multi-agent reinforcement learning algorithms that can improve single-agent learning, generalization, coordination, and human-AI collaboration. Natasha earned her PhD from MIT, where she focused on Affective Computing and new techniques for deep/reinforcement/machine learning. Natasha earned her Masters degree from the University of British Columbia, and undergraduate degrees in Computer Science and Psychology from the University of Regina.

Facebook event page: https://www.facebook.com/events/2990667311162405