Incoming computer science master’s students Alex Bie, Matt Regehr, Xinyu Shi and Yiping Wang have each received a Vector Scholarship in Artificial Intelligence from the Vector Institute. Valued at $17,500, these merit-based scholarships help recruit top students into AI-related master’s programs in an Ontario university.
Now in its fourth year, Vector Scholarships in Artificial Intelligence were awarded to 109 master’s students in this round of funding. As well as providing financial support, scholarship recipients have the opportunity to attend events and talks to meet leading AI researchers, clinicians, and teams from top employers, providing unique access to job opportunities. Scholarship recipients can also take advantage of Vector’s Digital Talent Hub, career development and networking events, and AI-specific career support programs.
“Congratulations to Alex, Matt, Xinyu and Yiping,” said Professor Justin Wan, Director of Graduate Studies at the Cheriton School of Computer Science. “These exceptional master’s students come to us with diverse research interests and backgrounds, but all four Vector Scholarship recipients are eager to apply their knowledge and experience to develop AI that improves our lives.”
Alex has held a variety of internships at which he has gained practical experience in AI systems, including work placements where he helped develop AI systems that allow data sharing and processing while guaranteeing data privacy. His goal is to build AI tools and to develop underlying theory that ensures AI systems integrate beneficially into our lives and societies.
Matt is interested in pushing the frontiers of AI through solid mathematical foundations. His primary research interests include statistical learning theory, reinforcement learning, and information theory. He particularly enjoys work that draws on several bodies of theory from measure theory and dynamical systems to functional analysis and mathematical statistics. His experience with research in biomedical imaging has also inspired his interest in privacy in machine learning.
Xinyu’s previous research projects have focused on using AI to help people better perceive the real-world and to interact effectively with virtual environments. She worked on generative models that increased the resolution and quality of pictures to bring old photographs back to life. She also proposed novel machine learning approaches that can recognize human actions and detect objects to support virtual and augmented reality.
The main challenge in AI systems that Xinyu has tackled is debugging neural networks. The space in which neural networks operate is hidden and often inexplicable, resulting in problems that are difficult if not impossible to solve when models perform unexpectedly. She is also concerned about bias in AI training data, which can lead to discriminatory AI and deep distrust in its decisions. But what if we developed visualizations that made AI’s functions clear and easy to understand? She hopes to design novel visualization systems that interact with models of AI systems that are interpretable, transparent and trustworthy.
After watching a 2015 TED lecture titled “How we’re teaching computers to understand pictures” by computer vision expert Fei-Fei Li, Yiping became captivated in understanding how computers extract and learn meaningful information from vast databases of photos. Given enough training data, computers can learn to reliably identify commonplace objects in photos — for example, a cat, tree or car — with great accuracy, but can they also be trained to analyze medical images and be used in clinical settings to assist medical professionals?
During his undergraduate degree, Yiping worked as a research intern at the University of British Columbia, where he proposed a simple but effective transfer learning algorithm to classify ovarian cancer histopathology — an algorithm that was able to outperform doctors without gynecological training. He also worked as a research intern at Imagia, a Montreal-based company that explores applications in computer vision and mapping of clinical information, where he helped develop a novel recurrent generative adversarial network algorithm to synthesize high-resolution 3D CT scans.
During his graduate studies, Yiping plans to enhance and expand his understanding of computer vision, especially segmentation. He is particularly enthusiastic about developing weakly supervised segmentation algorithms to segment cells in high-resolution histopathology images, and to combine single-cell sequencing to improve patient care and solve real-world clinical problems.
About the Vector Institute
The Vector Institute is an independent, not-for-profit corporation dedicated to research in the field of artificial intelligence, excelling in machine and deep learning. The Vector Institute launched in March 2017 with generous support from the Government of Canada, Government of Ontario, and private industry, and in partnership with the University of Toronto and other universities. The Vector Institute works with institutions, industry, start-ups, incubators and accelerators to advance AI research and drive its application, adoption and commercialization across Canada.
Vector Scholarship in Artificial Intelligence quick facts
- Number of master’s scholarships awarded in this cycle of funding: 109
- Number of scholarships awarded to date: 351
- Number of institutions with recipients in this cycle of funding: 13
- Number of AI programs receiving scholarships this cycle of funding: 34