julian.togelius.comJulian Togelius

Associate Professor Department of Computer Science and Engineering Tandon School of Engineering New York University 2 MetroTech Center, Brooklyn, NY 11201, USA Co-director of the NYU Game Innovation Lab Editor-in-Chief, IEEE Transactions on Games Contact: +1 917 285 3152 +46 705 192088 (Sweden)julian@togelius.com Research interests I’m working on artificial intelligence techniques for making computer games more fun, and on games for making artificial intelligence smarter. I ask what AI can do for games, and what games can do for AI. I want to make computer games adapt to their players through finding out what players want (whether they know it or not) and creating new game levels, challenges or rules that suit the players. Related to this is the challenge of making sense of large amounts of data generated by computer games, and on assisting human game designers in creating great game experiences. I also want to make opponents and collaborators in games more intelligent and believable, research that has applications far outside of computer games. I believe games, in particular video games, are perfect test-beds for AI methods. But it is important that you test your algorithms not just on a single game, but on a large number of games, so you focus on general intelligence and not just solving a single problem. That was the easily readable version of what I do. This is the more technical version: I’m interested in most topics in the intersection between computational intelligence and games. I currently focus on search-based procedural content generation, which means using evolutionary algorithms (or other global optimization algorithms) to search the space of game content, such as levels, maps or game rules. A key problem here is how to evaluate the candidate content, i.e. how to create a fitness function for game content. My and my colleagues attack this problem in a variety of ways, e.g. by creating theoretically founded and/or data driven models of player experience, by simulated gameplay using heuristic agents augmented by behavioural models of human players, and by measuring the performance of learning algorithms on candidate game rules. In order to do this, we need to be able to learn how to play a particular game as well as possible, and/or how to play it in a particular manner. These are interesting and worthy challenges in their own right, with additional applications to e.g. robotics. In this context, I’ve been working on evolutionary reinforcement learning in digital game domains, exploring techniques such as incremental evolution, competitive co-evolution and the hybridization of evolution with ontogenetic RL algorithms such as TD-learning. But it’s not enough to learn well; one must know how well one learns. My contribution to fair and reliable benchmarking of RL algorithms and game AI in general takes the form of inventing and running competitions based on digital action games, such as car racing and Super Mario Bros. Recently, I have been involved in starting and organizing the General Video Game Competition, and much of my efforts currently revolve around this framework. As mentioned above, we often want to learn not only how to play a game as well as possible, but also how to reproduce a particular playing style. We might also be interested in learning other aspects of the game player, notably what game experiences she prefers. (If we know how the player plays and what she wants, we can automatically evaluate new candidate game content before introducing it into the game.) Here, I have been working on player modelling through preference learning, and on data mining on large sets of game data. I’ve also been looking at various methods of reproducing player behaviour through machine learning based on player data. In the past, I have worked on several topics in evolutionary computation, in particular a project to unify evolutionary methods in a geometric framework. I’ve also at various points published research on e.g. evolutionary robotics and modularity in neural networks, and maintain an interest in these topics. And before getting caught up in computational intelligence, I was interested in consciousness and the problem of other minds, but I guess I’m too impatient for philosophy. Now, this text is already too long... Impact Google scholar citations page Selected media coverage Cleverness isn’t everything for a gaming artificial intelligence , New Scientist (18 February 2016). AI, acting and the weird future of open-world games , Eurogamer (21 November 2015). Why Artificial Intelligence Researchers Love ’Super Mario Bros.’ , VICE Motherboard (30 October 2015). Forskaren vill avdramatisera bilden av artificiell intelligens , Sydsvenskan/HD (14 February 2015) Malmöforskare kan göra speldesigners arbetslösa , Sydsvenskan/City (13 January 2015) Gaming bots learn from how we play , New Scientist (24 September 2014) AI card game knows how to bend the rules , New Scientist (3 May 2013) Game do futuro "sentirá" jogador, diz pesquisador , O Globo (2012) Online games are a gold mine for design ideas ( Local copy ), New Scientist (23 August 2010) Des jeux vidéo sur mesure , Le Monde (9 December 2009) Should Computer Games Adapt to the Way You Play? , Slashdot (13 October 2009) Adaptive games promise high scores for everyone ( Local copy ), New Scientist (7 October 2009) AI tested on ’Super Mario’ video game , MSNBC / Discovery Channel News Service (17 August 2009) Quand c’est l’ordinateur qui joue à Mario , Le Monde (7 August 2009) Race is on to evolve the ultimate Mario , New Scientist (5 August 2009) Mario AI Competition , Slashdot (5 August 2009) Game that develops itself could save industry , The Guardian (21 January 2009) Autonomous driving systems aim to drive dirty , New Scientist (23 March 2007) "Another Step Towards the Driverless Car" , Slashdot (23 March 2007) Ongoing organisational activities General chair for IEEE Computational Intelligence and Games 2017. Keynotes chair for Foundations of Digital Games 2017. Co-organizer of the General Video Game Playing Competition . Past organisational activities Tutorials and keynotes chair of IEEE Computational Intelligence and Games 2016. Track chair for the DETA track of GECCO 2016. Program chair for Foundations of Digital Games 2015. Track chair for the DETA track of GECCO 2014. Technical co-chair for IEEE Conference of Computational Intelligence and Games 2013. Workshops chair for Artificial Intelligence and Interactive Digital Entertainment 2013. Workshops chair for Foundations of Digital Games 2013. Associate Editor for IEEE Transactions on Computational Intelligence and Games Co-organiser of the Dagstuhl seminar on Artificial and Computational Intelligence in Games . Competitions chair for IEEE CIG 2012. Co-chair of EvoGames (part of EvoApplications) 2012. Co-chair of the Procedural Content Generation Workshop at FDG 2012. Chair of IEEE CIS Games TC 2011-2012 Competitions chair for IEEE CIG 2011. Co-chair of EvoGames (part of EvoApplications) 2011. Co-organizer of the Procedural Content Generation Workshop at FDG 2011. General co-chair for CIG 2010 Competitions chair for ICE-GIC 2010 Co-organizing the 2010 Mario AI Championship , including the world’s first PCG competition! Co-organizing the 2009 Mario AI Competition Publicity Chair for IEEE CIG 2009 Competitions Chair for ICE-GIC 2009 Co-organizing the Simulated Car Racing Championship 2009 Co-organizing the CEC 2009 Special Session on Computational Intelligence and Games CIG 2008 Simulated Car Racing Competition WCCI 2008 Simulated Car Racing Competition CEC 2007 Simulated Car Racing Competition CIG 2007 Simulated Car Racing Competition Current and past PhD Students Michael Green (started 2016, NYU): AI and games. Athoug Alsoughayer (started 2016, NYU): AI and games. Ahmed Khalifa (started 2015, NYU): General level generation. Andre Mendes (started 2015, NYU): Hyper-heuristic game playing. Tiago Machado (started 2015, NYU):...