CIFAR: The Canadian Institute That Funded the Deep Learning Revolution

When the AI research community abandoned neural networks in the 1990s, one Canadian institution kept funding the work. That decision led to deep learning, the technology behind ChatGPT, autonomous vehicles, and the current AI revolution. The institution was CIFAR—the Canadian Institute for Advanced Research.

Today, CIFAR administers Canada's $2.4 billion national AI strategy and has helped train over 1,500 AI researchers. Three of its fellows—Geoffrey Hinton, Yoshua Bengio, and Yann LeCun—won the Turing Award for their work on deep learning.

What Is CIFAR?

CIFAR (Canadian Institute for Advanced Research) is a Toronto-based research organization that supports global networks of researchers working on complex scientific and social challenges. Unlike traditional research institutes, CIFAR operates as a "virtual university"—connecting researchers across institutions rather than housing them in a single location.

CIFAR's model emphasizes long-term, high-risk research that may not produce immediate results but can lead to transformative discoveries. This patient approach proved essential for supporting neural network research during the "AI winter" of the 1990s and 2000s.

How Was CIFAR Founded?

The idea for CIFAR originated with John Leyerle, a professor of English and dean of the School of Graduate Studies at the University of Toronto, who began advocating for an advanced studies institute in 1978.

Founding Vision

Leyerle envisioned an institute that would "foster basic, conceptual research of high quality at an advanced level across the full spectrum of knowledge in the humanities, social sciences, natural sciences and life sciences."

J. Fraser Mustard, a medical doctor and researcher in early childhood development, was appointed as CIFAR's founding president in January 1982. He conceived of CIFAR as a "virtual university" that prioritized long-term, collaborative inquiry over short-term outputs.

Early Support

The early commitment to AI research—establishing an AI program in 1983, just one year after founding—positioned CIFAR at the center of what would become the deep learning revolution.

How Did CIFAR Support the Deep Learning Pioneers?

CIFAR's most consequential decision was supporting neural network research when most of the AI community had moved on to other approaches.

Geoffrey Hinton Joins CIFAR (1987)

Geoffrey Hinton first joined CIFAR in 1987 as a Fellow in the Artificial Intelligence, Robotics & Society program. He had moved from the United States to Canada partly due to disillusionment with Reagan-era politics and disapproval of military funding for AI research.

"Their determination to follow paths others had rejected is testament to the importance of supporting basic research, which may not always produce results quickly – but can lead to profound discoveries over time." — Meric Gertler, President, University of Toronto, on the Turing Award winners

With financial support from CIFAR and NSERC (Natural Sciences and Engineering Research Council of Canada), Hinton continued developing neural networks during the "AI winters"—periods when funding and interest in AI research declined sharply.

Neural Computation and Adaptive Perception (2004)

In 2004, Hinton and collaborators including Yoshua Bengio and David Fleet proposed a new CIFAR program: Neural Computation and Adaptive Perception (NCAP). This program—now called Learning in Machines & Brains—became the incubator for modern deep learning.

The program brought together researchers who believed in neural network approaches when mainstream AI research focused on symbolic methods and expert systems.

The Turing Award Trio

Beginning in 2004, Hinton, Bengio, and LeCun began meeting regularly through CIFAR, sharing ideas that informed each other's work. Their collaboration produced breakthroughs in deep neural networks that earned them the 2018 Turing Award—often called the "Nobel Prize of Computing."

In 2024, Hinton was awarded the Nobel Prize in Physics for his work on artificial neural networks—further validating CIFAR's decades-long investment in the field.

What Is the Pan-Canadian AI Strategy?

In 2017, the Government of Canada asked CIFAR to develop and lead the Pan-Canadian Artificial Intelligence Strategy—the world's first national AI strategy.

Phase 1 (2017)

The strategy established three national AI institutes, each reflecting regional academic strengths:

• Vector Institute (Toronto): Enterprise AI, health, finance
• Mila (Montreal): Deep learning fundamentals, AI safety
• Amii (Edmonton): Reinforcement learning, machine learning applications

Phase 2 (2022)

Phase 2 focused on translating AI research into commercial applications while continuing to attract and retain top talent.

Budget 2024: $2.4 Billion

Budget 2024 proposed a $2.4 billion investment to secure Canada's AI advantage—the largest federal AI commitment to date. This funding supports:

• Continued development of national AI institutes
• AI safety research through CAISI (Canadian AI Safety Institute)
• Compute infrastructure and sovereignty
• Talent development and international recruitment

What Is the Learning in Machines & Brains Program?

The Learning in Machines & Brains program is CIFAR's flagship AI research initiative—the direct descendant of the Neural Computation and Adaptive Perception program that launched the deep learning revolution.

Program Focus

The program aims to understand the computational and mathematical principles that enable intelligence through distributed learning in neural networks—whether in biological brains or artificial systems.

Impact

Research from Learning in Machines & Brains has:

• Pioneered deep learning techniques now used in computer vision, speech recognition, and natural language processing
• Influenced products at Google, Meta, Microsoft, and other major tech companies
• Trained researchers who now lead AI labs worldwide
• Produced two Turing Award winners (Bengio, LeCun) and one Nobel laureate (Hinton)

DLRL Summer School

The Deep Learning and Reinforcement Learning (DLRL) Summer School is a signature program that trains next-generation AI researchers. Hosted annually in partnership with Vector, Mila, and Amii, DLRL brings together students from around the world to learn from leading researchers.

What Are CIFAR's Other Research Programs?

CIFAR supports 15 research programs across diverse fields, reflecting its founding vision of advancing knowledge "across the full spectrum."

Current Programs

AI Catalyst Grants

CIFAR's AI Catalyst Grants fund high-risk, high-reward research projects. In 2024-2025, seven new projects received funding of up to $50,000 per year for collaborative research between Canada CIFAR AI Chairs and other researchers.

What Is the Canada CIFAR AI Chairs Program?

The Canada CIFAR AI Chairs program recruits and retains world-class AI researchers at Canadian institutions.

Program Stats

The program has been remarkably successful at attracting international talent. Half of the recruited researchers came from outside Canada, drawn by the combination of research funding, institutional support, and Canada's welcoming immigration policies.

Notable Chair Holders

Many Canada CIFAR AI Chairs have gone on to lead major research initiatives, found companies, or take leadership positions at global tech firms—creating a multiplier effect for Canada's AI ecosystem.

How Is CIFAR Funded?

CIFAR receives funding from multiple sources, reflecting its unique position between government, academia, and the private sector.

Funding Sources

Key Government Investments

What's Next for CIFAR?

Current Priorities

1. AI Safety Research: CIFAR hosts the Canadian AI Safety Institute (CAISI), funding research on safe and beneficial AI
2. Talent Development: Continuing to train the next generation through chairs, fellowships, and summer schools
3. Commercial Translation: Helping move research from labs to industry applications
4. International Collaboration: Expanding global research networks

The CAISI Research Program

In June 2025, CIFAR announced funding for 10 new AI safety research projects through the Canadian AI Safety Institute Research Program—reflecting growing emphasis on ensuring AI systems are safe and beneficial.

---

Frequently Asked Questions

What does CIFAR stand for?

CIFAR stands for the Canadian Institute for Advanced Research. It's a Toronto-based research organization that supports global networks of researchers working on complex challenges across science and society.

When was CIFAR founded?

CIFAR was founded in 1982 by Dr. J. Fraser Mustard. The idea originated with John Leyerle of the University of Toronto in 1978.

How did CIFAR help create deep learning?

CIFAR provided long-term funding and a collaborative environment for neural network researchers during the "AI winter" when most institutions had abandoned the field. In 2004, Geoffrey Hinton launched the Neural Computation and Adaptive Perception program at CIFAR, bringing together researchers including Yoshua Bengio and Yann LeCun. Their collaboration produced breakthroughs in deep learning that earned them the 2018 Turing Award.

What is the Pan-Canadian AI Strategy?

The Pan-Canadian AI Strategy is Canada's national AI strategy, launched in 2017 and administered by CIFAR. It was the world's first national AI strategy. The strategy has invested over $2.4 billion in AI research, talent development, and the establishment of three national AI institutes: Vector (Toronto), Mila (Montreal), and Amii (Edmonton).

What is the Learning in Machines & Brains program?

Learning in Machines & Brains is CIFAR's flagship AI research program, co-directed by Yoshua Bengio and Yann LeCun. It evolved from the Neural Computation and Adaptive Perception program founded by Geoffrey Hinton in 2004. The program aims to understand intelligence in both biological brains and artificial systems.

How many researchers does CIFAR support?

CIFAR supports over 400 researchers from 21 countries and more than 140 institutions. Through the Pan-Canadian AI Strategy, over 100 Canada CIFAR AI Chairs have been recruited, and more than 1,500 graduate students have been trained.

Who leads CIFAR?

Stephen Toope has served as CIFAR's President and CEO since November 2022. Irfhan Rawji is the chair of CIFAR's Board of Directors.

---

Related Reading

The Deep Learning Pioneers

• Geoffrey Hinton: Godfather of AI — The Nobel laureate who led CIFAR's neural network program
• Yoshua Bengio: Deep Learning Pioneer — Turing Award winner and Learning in Machines & Brains co-director
• Rich Sutton: Father of Reinforcement Learning — The researcher who shaped AI at University of Alberta

Canada's AI Institutes

• Vector Institute Complete Guide — Toronto's AI research institute, established through CIFAR's AI Strategy
• Mila Complete Guide — Montreal's AI powerhouse, co-founded by Yoshua Bengio
• Amii Complete Guide — Edmonton's reinforcement learning center

AI Safety

• CAISI: Canada's AI Safety Institute — The safety research initiative hosted by CIFAR

The Canadian AI Ecosystem

• The AI Triangle: Canada's Three Superpowers — How Toronto, Montreal, and Vancouver collaborate
• Canadian AI Unicorns: Complete Guide — Companies built on Canadian AI research

---

Join AGI House Canada to connect with Canada's AI research community. Scan the QR code to join our WhatsApp group or subscribe to our newsletter.

Data current as of January 2026. Funding data from Government of Canada announcements and CIFAR reports. Historical information from CIFAR, Wikipedia, and ACM Turing Award documentation.