events

MARL Chapter 9.9: Population-Based Training

A recurring online study group works through Chapter 9 of the textbook Multi-Agent Reinforcement Learning: Foundations and Modern Approaches, continuing its discussion of population-based training and the double oracle algorithm in zero-sum

Starts

2026-09-22T00:30:00+00:00

Organizer

Silicon Valley Generative AI ~ The AI Collective Network

Location

Online event

A recurring online study group works through Chapter 9 of the textbook Multi-Agent Reinforcement Learning: Foundations and Modern Approaches, continuing its discussion of population-based training and the double oracle algorithm in zero-sum games. This session looks at the algorithm's theoretical limits against exact solution methods for tabular stochastic games, starting from minimax solutions and value iteration in the zero-sum case. If time allows, the group moves to general-sum games and a meta-solver such as WoLF-PHC. Organised by the Silicon Valley Generative AI meetup group, which posts notes, slides and recordings.