Towards Ad-Hoc Teamwork for Improved Player Experience | Sam Devlin

Towards Ad-Hoc Teamwork for Improved Player Experience (opens in new tab)
ICARL Seminar Series – 2022 Winter
Seminar by Sam Devlin

Abstract:
Collaborative multi-agent reinforcement learning research often makes two key assumptions: (1) we have control of all agents on the team; and (2) maximising team reward is all you need. However, to enable human-AI collaboration, we need to break both of these assumptions. In this talk I will formalise the problem of ad-hoc teamwork and present our proposed approach to meta-learn policies robust to a given set of possible future collaborators. Then talk about recent work on modelling human play, showing reward maximisation may not be sufficient when trying to entertain billions of players worldwide.

——————————————————
Links
Sam Devlin
Site: aka.ms/samdevlin
Twitter: x.com/smdvln

ICARL
Site: icarl.doc.ic.ac.uk
Twitter: x.com/ic_arl
YouTube: youtube.com/ICARLSeminars
——————————————————

Intro and Outro music courtesy of Bensound.com – Funky Suspense by Benjamin Tissot

Date:
Haut-parleurs:
Sam Devlin