Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Industrial Research Chairs

Université de Montréal

$173,625.00

NSERC - Ubisoft Industrial Research Chair on Learning Representations for Immersive Video Games

FY 2010-11

Summary
Artificial Intelligence research has been heavily driven by the availability of large quantities of data and by the discovery of machine learning algorithms that can exploit it. Algorithms for learning new data representations from large mostly unlabeled datasets have progressed significantly in recent years, and are finding applications in many areas. Models with several levels of representation are called deep. Our long term research goal is to develop deep models of high dimensional dynamic processes such as you would find in video, music, animations etc, training these models from very large quantities of data for which very little human labeling is necessary. This project aims at creating applications of these new approaches in the video game industry, in collaboration with partner Ubisoft. The aim is to exploit these advances to enhance the immersion and the fun experienced by game players. This will be done by exploiting learning algorithms to develop better technology for adapting the player's game experience to the player's style, for generating richer video content, and to empower artists to more efficiently create a larger variety of credible virtual environments. This includes modeling characters and players, taking into account the 3-dimensional, sequential, and behavioral aspects of video games. For example, by characterizing the style, mood, and skill of players we can better match them with each other and with non-player characters, as well as present them with better fitting challenges and environments. By improving the models of animated characters, we can help game developers compress more animations into their games, avoid awkward configurations of the characters skeletons, better blend one configuration into the next, and generate a varied set of movements instead of always repeating the same canned sequences, and generally produce more, and more fun game content, faster and better.
Program
Industrial Research Chairs
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
335231-2010
Name nom
Bengio, Yoshua
Department département
Informatique et recherche opérationnelle
Organization id
63
Institution établissement
Université de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2010
Competition year année de concours
2010
Award amount
173625
Program id
IRCPJ
Program name en
Industrial Research Chairs
Program name fr
Professeurs-chercheurs industriels
Group en
Research partnerships programs
Group fr
Programmes de partenariat
Committee code
141
Committee name en
University/Industry
Committee name fr
Universités/industrie
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2805
Research subject group en
Artificial intelligence
Research subject group fr
Intelligence artificielle
Research subject en
Learning and inference theories
Research subject fr
Théories de l'apprentissage et l'inférence
Partie
6
Nb partie
10
Application title
NSERC - Ubisoft Industrial Research Chair on Learning Representations for Immersive Video Games
Keyword
NSERC
Application summary
Artificial Intelligence research has been heavily driven by the availability of large quantities of data and by the discovery of machine learning algorithms that can exploit it. Algorithms for learning new data representations from large mostly unlabeled datasets have progressed significantly in recent years, and are finding applications in many areas. Models with several levels of representation are called deep. Our long term research goal is to develop deep models of high dimensional dynamic processes such as you would find in video, music, animations etc, training these models from very large quantities of data for which very little human labeling is necessary. This project aims at creating applications of these new approaches in the video game industry, in collaboration with partner Ubisoft. The aim is to exploit these advances to enhance the immersion and the fun experienced by game players. This will be done by exploiting learning algorithms to develop better technology for adapting the player's game experience to the player's style, for generating richer video content, and to empower artists to more efficiently create a larger variety of credible virtual environments. This includes modeling characters and players, taking into account the 3-dimensional, sequential, and behavioral aspects of video games. For example, by characterizing the style, mood, and skill of players we can better match them with each other and with non-player characters, as well as present them with better fitting challenges and environments. By improving the models of animated characters, we can help game developers compress more animations into their games, avoid awkward configurations of the characters skeletons, better blend one configuration into the next, and generate a varied set of movements instead of always repeating the same canned sequences, and generally produce more, and more fun game content, faster and better.