The Vector Institute held its third annual Remarkable conference on February 19–20 in Toronto, drawing more than 1,500 researchers and industry participants in person and online, according to the institute’s account of the event. The second day centered on a poster session in which 60 research projects from across the Vector network were presented in the main auditorium.
The institute described three recurring themes across the 60 presentations: health care AI, AI safety and responsible AI, and foundation model architecture.
Health care AI accounted for close to 20 percent of the presentations, the institute said, with projects covering voice-controlled surgical navigation, cancer detection using foundation models, clinical decision support systems, and medical imaging. AI safety work included research on bias detection across demographics and dialects, privacy preservation, fairness optimization, and methods for identifying and reducing AI vulnerabilities. Foundation model work addressed training efficiency, multimodal capabilities, and new architectures for language and vision systems.
The institute described the portfolio as evidence of “Canadian AI leadership extending from Toronto-based optimization startups commercializing breakthrough technologies to Vector’s comprehensive medical AI pipeline positioning Canada as a global hub for healthcare innovation.” That characterization was part of the institute’s own summary of the event, not an independent assessment.
Vector was founded in 2017 as a not-for-profit research institute focused on machine learning and deep learning. Its annual Remarkable conference, now in its third year, is aimed at connecting academic research with industry application. The institute does not take equity stakes in the startups it supports, distinguishing it from incubators that operate on that model.
The 2026 poster session covered ground across the institute’s research community, which includes faculty affiliates, postdoctoral researchers, and industry-funded chairs. The institute publishes abstracts and research summaries through a downloadable poster booklet; the full collection was organized by research theme corresponding to the February 20 presentations.
No external independent assessments of the research quality or outcomes from the poster session were available in the source material.