Reasoning about Causality in Games

11^th October 2017, 13:00
Piotr Krysta
University of Liverpool

Abstract

Causal reasoning and game-theoretic reasoning are fundamental topics in many disciplines, yet despite this, a formal framework that supports both these forms of reasoning has, until now, been lacking. In this talk, we offer a solution in the form of causal games, which can be seen as extending Pearl’s causal hierarchy to the game-theoretic domain, or as extending Koller and Milch’s multi-agent influence diagrams to the causal domain. In doing so, we show how the (causal) dependencies in strategic decision-making scenarios – either between variables, or between strategies – can be modelled in a uniform, principled manner, and how to generalise conditional, interventional, and counterfactual queries to games. We also briefly sketch some of our earlier game-theoretic results in these models, connections to other formalisms, and possible domains of application. Finally, we will provide an overview of our codebase PyCID, which can be used to implement causal games.

Biography

ames Fox is a DPhil candidate in Computer Science at the University of Oxford, supervised by Alessandro Abate and Michael Wooldridge. He is motivated by problems that aim to build safe, scalable, and beneficial AI systems; in particular, his research primarily focuses on connecting ideas from game theory, control, and machine learning to understand behaviour in multi-agent systems. James has a master's degree in physics from the University of Cambridge, and he has also worked for several start-ups and with government.

Lewis Hammond is based at the University of Oxford where he is a DPhil candidate in computer science and a DPhil affiliate at the Future of Humanity Institute. He is also currently serving as the acting executive director of the Cooperative AI Foundation, and is affiliated with both the Centre for the Governance of AI and the Future Society. His research concerns safety, control, and incentives in multi-agent systems and in practice spans game theory, formal methods, and machine learning. Before coming to Oxford he obtained a BSc (Hons) in mathematics and philosophy from the University of Warwick, and an MSc in artificial intelligence from the University of Edinburgh.