November 3, 2018

Machine Learning and Games: IDG published in IEEE Transactions on Evolutionary Computation Journal

  • Tagged:
  • Student Work

Machine Learning in the Gaming Industry: Cutting Edge Research at the IDG

The Institute of Digital Games is excited to announce that the work of Daniele Gravina, Dr Antonios Liapis, and Prof. Georgios Yannakakis on "Quality Diversity Through Surprise" will be published IEEE Transactions on Evolutionary Computation Journal. The journal has an Impact Factor of 8.124 and is ranked 1st Machine Learning & Artificial Intelligence and also ranked third by Guide2Research in Computer Science and Electronics in general.

The paper, “Quality Diversity Through Surprise,” investigates the use of surprise as a mechanism of search finding that it turns out to be highly beneficial for the efficiency and robustness of quality diversity (QD) algorithms. The surprise-based QD algorithms introduced in this IEEE TEC paper were tested thoroughly across several procedurally generated maze navigation tasks. Overall our findings suggest that if surprise and novelty operate synergistically for divergence they yield quality diversity algorithms of improved efficiency, speed and robustness. The source code and more information about surprise search can be found on the AutoGame website.

Games that use learning AI are also explored on the AutoGame website such as a paper and source code covering a Computational Game Designer that sees the balance of the level.

Learning AI in Video Games

The Institute of Digital Games – University of Malta was ranked among the top 25 post-graduate game design programs by the Princeton Review in 2017 and 2018 and is proud to continue publishing impactful research in areas related to game design and artificial intelligence.