Postdoctoral Research Assistant at the University of Maastricht in the Department of Data Science and Knowledge Engineering (DKE) , I work mainly in the Digital Ludeme Project, a five year ERC-funded research project (European Research Council) from April 2018 to March 2023.
This project is a computational study of the world's traditional strategy games throughout recorded human history. It aims to improve our understanding of traditional games using modern AI techniques, to chart their historical development and explore their role in the development of human culture and the spread of mathematical ideas.
The development of games goes hand in hand with the development of human culture. Games offer a rich window of insight into our cultural past, but early examples were rarely documented and our understanding of them is incomplete. While there has been considerable historical research into games and their use as tools of cultural analysis, much is based on the interpretation of partial evidence with little mathematical analysis. This project will use modern computational techniques to help fill these gaps in our knowledge empirically.
The project will represent games as structured sets of ludemes (units of game-related information), which will allow the full range of traditional strategy games to be modelled in a single software system for the first time. This system will not only model and play games, but will evaluate reconstructions for quality and authenticity, and automatically improve them where possible. This will lay the foundations for a new field of study called Digital Archaeoludology (DA), which will involve addressing technical challenges that could yield significant benefits in their own right, particularly in Artificial Intelligence (AI) and General Game Playing.
The ludemic model (LUDII) reveals innate mathematical relationships between games, allowing phylogenetic analysis. This provides a mechanism for creating a family tree/network of traditional games, which could reveal missing links and allow ancestral state reconstruction to shed light on the gaps in our partial knowledge. Locating ludemes culturally provides a mechanism for creating interactive maps that chart the transmission of mathematical ideas across cultures through play. This project :
In parallel, I continue my current research based on my thesis entitled "A Stochastic Constraint-Based Approach to General Game Playing". Indeed, my main theme is the General Game Playing (GGP) whose the primary objective is to create an artificial intelligence which is able to play strategically at a variety of games. Thanks to my work, I introduced a new kind of approach based on the stochastic constraint satisfaction problem (SCSP) and on the Bandit-based stochastic sampling. One of these key characteristics is to simplify the resolution of each problem by demonstrating the equivalence between the detected constraint/solution symmetries in the constraint network generated with the structural/strategies symmetries described by the GGP games. In practical, this work is substantiated by the general game player called WoodStock which won the International General Game Playing Competition 2016 (IGGPC'16).
More information about my thesis here.
Keywords : General Game Playing, machine learning, knownledge representation, constraint programming.
Artificial Intelligence is no substitute for natural stupidity.