Skip to main content

Faculty of Business and Information Technology Project Summaries

Supervisors  

Amirali Abari | Andrew Hogue | Pooria Madani

 

Supervisor name: Amirali Abari

Project title: Graph Neural Networks for Recommender Systems

Summary of research project: Recommender systems are prevalent in our day-to-day life. They intelligently recommend desirable options to us (e.g., books in Amazon, movies in Netflix, etc.), which are consistent to our own taste or preferences. Several recent developments have been made in applying deep learning in recommender systems. Of important and promising direction is utilizing graph embedding and graph convolutional networks for empowering recommender systems with graph-structured side information (e.g., social networks). Our goal is to develop effective deep learning algorithms for large-scale recommender systems in the presence of graph-structured side information.

Student responsibilities/tasks:

  • The Student is expected to review the relevant research literature along with other team members.
  • The Student also will be heavily involved in development of the AI technologies and will co-author the consequent paper written from their involvement in this Project.
  • This Project provides a unique opportunity to the candidate for fostering their knowledge in AI, graph neural networks, and deep learning.

Student qualifications required:

  • The Student is expected to have familiarity with PyTorch and Deep Learning and be excellent in programming in Python.

Expected training/skills to be received by the Student:

  • Graph Neural Networks
  • Programming in PyG
  • Literature Review
  • Academic Writing

Length of award: 16 Weeks

Location of award: In-Person

Available Award: NSERC USRA or Ontario Tech STAR

 

Supervisor name: Andrew Hogue

Project title: Interactive 4D Avatars for Games

Summary of research project: In this project, students will explore the space of interactive 4D realistic avatars. They will explore emerging techniques and technologies to record and capture photorealistic 3D recordings of performances in our volumetric capture studio and evaluate the use of these avatars for Games and XR applications.

Student responsibilities/tasks:

  • Students will be tasked with learning about Volumetric Video, Gaussian Splatting, and Gaussian Avatars and to work with other students on using existing tools and machine learning algorithms to train a set of Gaussian avatars for use in a research study on the use of photorealistic avatars in interactive scenarios.

Student qualifications required:

  • Students must be motivated to learn new skills related to artificial intelligence and machine learning.
  • Open to Computer science and/or IT/Game Dev/Engineering students.
  • Introductory knowledge of Machine Learning or motivation to learn ML techniques are desired.
  • Basic Python skills are necessary.
  • Unity/Unreal Engine knowledge is preferred.

Expected training/skills to be received by the Student:

  • Knowledge and expeirence with Volumetric Video tools.
  • Knowledge and experience creating Gaussian Splats
  • Practicing python and machine learning skills
  • Practicing Problem Solving and Critical Thinking skills
  • Experience with Unity/Unreal Engine

Length of award: 16 Weeks

Location of award: In-Person

Available Award: NSERC USRA or Ontario Tech STAR

 

Supervisor name: Pooria Madani

Project title: Federated Learning Machine Learning

Summary of research project: This project aims to develop a simulation environment tailored for testing and evaluating federated learning algorithms in satellite constellations. Federated learning is a distributed machine learning approach where models are trained across multiple devices or nodes, such as satellites, without centralized data aggregation. This is particularly relevant for satellite constellations, where communication bandwidth is limited, and data privacy is crucial.

Student responsibilities/tasks:

  • Be proficient in Python to develop full orbital simulations of satellite constellations.
  • Implement and visualize various orbital configurations, ensuring accurate simulation of satellite motion.
  • Research existing work in satellite constellations and related algorithms.
  • Document all work, including code and findings, on the research lab's wiki for future reference.

Student qualifications required:

  • Strong understanding of Python programming.
  • Comfortable with geometry and mathematics, especially implementing mathematical formulas in Python.
  • Experience with scientific Python libraries (e.g., NumPy, SciPy, Matplotlib).
  • Familiarity with machine learning is a bonus.

Expected training/skills to be received by the Student:

  • Gain sufficient knowledge of orbital mechanics to effectively build the simulation environment.
  • Acquire a solid understanding of backpropagation-based learning algorithms to effectively test them in a federated learning environment.

Length of award: 16 Weeks

Location of award: Hybrid

Available Award: NSERC USRA or Ontario Tech STAR