Canada Spends Logo

Explore our Budget 2026 analysis

Grant: Discovery Grants Program - Individual

Université du Québec à Montréal

$27,000.00

A unified framework for image analysis and generation by integrating machine learning to functional models

FY 2023-24

Summary
Images and videos form a new unique type of data and I seek to provide models, tools and software for the analysis of these. My Discovery research program aims at developing novel methodologies for image analysis and generation by combining functional models, space shape analysis and machine learning through probabilistic graphical models. While the long-term goal of my research program is to establish a new family of distribution suited for prediction, inference and sampling of spatiotemporal functional data, I propose to achieve this broad goal by first working towards three short-term objectives.  By integrating neural networks to functional data analysis models I seek to define distributions for spatiotemporal data in a new way. The machine learning component of what we propose provides us with the needed tools to handle the high dimension of the data while the functional data methods provide us with the interpretation we need to disseminate our results to a broad audience. Consequently, my first short-term objective is to develop a large class of functional neural networks. My students and I have first developed a neural network for functional responses and we now plan to develop a more general network architecture starting with a functional autoencoder. I am also looking to extend probabilistic graphical models to consider the structural and geometric aspect of landmarks and the curves within images. Probabilistic graphical models are flexible enough to consider distributions with various characteristics but are not yet suited for the curves obtained when processing images with edge-detection algorithms. Extending those to consider the shape of objects in images through the lens of functional data analyses would be a significant step towards the definition of a novel family of models for both the analysis and the sampling of new images. Thus, my second short-term goal is to establish new objective functions needed to train functional shape generative models. A third short-term objective within my research program is to stay connected with applications for the family of proposed models. In neurosciences, brain scans such as magnetic resonance imaging, are essential to deepen our understanding of neurological disease and the models I propose allow for inference on these images. Staying connected with applications is a way to motivate further model development and to better understand the strengths and the weaknesses of the models I propose. An interesting application for the generative aspect of my proposed theory would the democratization of art assets' creation in multimedia content. What I propose provides an inexpensive alternative to procedural content generation for the generation of meshes, skyboxes, landscapes and humanoid models in multimedia content. Consequently, my Discovery research will provide necessary tools to visual artists, neurologists and computer vision engineers alike.
Program
Discovery Grants Program - Individual
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data

As published

Application id
RGPIN-2023-05155
Name nom
Beaulac, Cédric
Department département
Département de Mathématiques
Organization id
57
Institution établissement
Université du Québec à Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2023
Competition year année de concours
2023
Award amount
27000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1100
Area of application group en
Health, education and social services
Area of application group fr
Santé publique, éducation et services sociaux
Area of application en
Health, education and social services
Area of application fr
Santé publique, éducation et services sociaux
Research subject code
9999
Research subject group en
Not available
Research subject group fr
Not available
Research subject en
Non disponible
Research subject fr
Non disponible
Application title
A unified framework for image analysis and generation by integrating machine learning to functional models
Keyword
Computational Statistics; Dimension Reduction; Functional Data Analysis; Image Analysis; Image Generation; Probabilistic Graphical Models; Variable Importance Analysis
Application summary
Images and videos form a new unique type of data and I seek to provide models, tools and software for the analysis of these. My Discovery research program aims at developing novel methodologies for image analysis and generation by combining functional models, space shape analysis and machine learning through probabilistic graphical models. While the long-term goal of my research program is to establish a new family of distribution suited for prediction, inference and sampling of spatiotemporal functional data, I propose to achieve this broad goal by first working towards three short-term objectives.  By integrating neural networks to functional data analysis models I seek to define distributions for spatiotemporal data in a new way. The machine learning component of what we propose provides us with the needed tools to handle the high dimension of the data while the functional data methods provide us with the interpretation we need to disseminate our results to a broad audience. Consequently, my first short-term objective is to develop a large class of functional neural networks. My students and I have first developed a neural network for functional responses and we now plan to develop a more general network architecture starting with a functional autoencoder. I am also looking to extend probabilistic graphical models to consider the structural and geometric aspect of landmarks and the curves within images. Probabilistic graphical models are flexible enough to consider distributions with various characteristics but are not yet suited for the curves obtained when processing images with edge-detection algorithms. Extending those to consider the shape of objects in images through the lens of functional data analyses would be a significant step towards the definition of a novel family of models for both the analysis and the sampling of new images. Thus, my second short-term goal is to establish new objective functions needed to train functional shape generative models. A third short-term objective within my research program is to stay connected with applications for the family of proposed models. In neurosciences, brain scans such as magnetic resonance imaging, are essential to deepen our understanding of neurological disease and the models I propose allow for inference on these images. Staying connected with applications is a way to motivate further model development and to better understand the strengths and the weaknesses of the models I propose. An interesting application for the generative aspect of my proposed theory would the democratization of art assets' creation in multimedia content. What I propose provides an inexpensive alternative to procedural content generation for the generation of meshes, skyboxes, landscapes and humanoid models in multimedia content. Consequently, my Discovery research will provide necessary tools to visual artists, neurologists and computer vision engineers alike.