A collaboration to turn a technological and creative corner
Ubisoft wishes to turn a new technological and creative corner by enabling the creation of games systems capable of evolving using the real-time analysis of the players' behaviour and to offer each player a personalised experience. The Flowers* and Mnemosyne project teams are providing their expertise in order to address these challenges faced by Ubisoft.
Machine learning to analyse game sequences
The Mnemosyne and Flowers project teams are joining forces to develop new machine learning techniques in order to analyse bodies of data representing Ubisoft game sequences. The aim is to see if any parts of the solution can be provided to the following questions using these data:
- When sequences are similar, can stereotypical phases in the game be identified?
- When similar sequences evolve slowly, can the improvement in a player's skills be measured?
- When the evolutions are different, can player profiles be distinguished? If so, can these profiles be qualified?
- Will it be possible to use all of the knowledge extracted in order to improve the player's experience according to his/her profile?
Intrinsically motivated exploration to test virtual worlds
The Flowers project team's idea is to study to what extent intrinsically motivated exploration algorithms can enable the effective exploration of virtual environments and to learn to produce a variety of effects within them. From a fundamental perspective, this means studying how to scale up the artificial curiosity algorithms the teams have been working on and which, until now, have been tested on robots.
In applicative terms, the purpose would be to use these algorithms as tools enabling video games under construction to be tested. For example, when a developer modifies a virtual world by adding new opportunities for interaction, this would make it possible to verify that the players find the solution and/or that this does not cause any unexpected behaviours.