Chris Maddison | Blood from a Stone: Finding Signal when Data is Scarce & Experiments are Expensive скачать в хорошем качестве

Chris Maddison | Blood from a Stone: Finding Signal when Data is Scarce & Experiments are Expensive

Discover how to extract maximum value from limited data when experiments cost millions. Chris Maddison, Vector Faculty Member and founding AlphaGo team member, tackles AI's most critical bottleneck: learning when data is scarce and verification is expensive. Why This Matters: Large language models consume data faster than new information is produced. In fields like drug discovery, testing a single candidate costs hundreds of millions of dollars. This talk reveals methodologies to learn effectively despite these constraints—the key to unlocking AI's potential in medicine, scientific research, and complex real-world applications. What You'll Learn: Why data scarcity is AI's most important unsolved problem How expensive verification bottlenecks breakthroughs in drug discovery and medicine Techniques to refine existing datasets and extract hidden value Gradient estimation methods now standard in deep learning Practical approaches from cutting-edge research addressing this challenge About the Speaker: Chris Maddison is an Assistant Professor at the University of Toronto, a Canada CIFAR AI Chair, and Vector Faculty Member. He designs machine learning algorithms that make accurate predictions in complex settings like medicine and drug discovery. Chris is renowned for gradient estimation techniques that became foundational tools in deep learning and for his role as a founding member of the AlphaGo project—the first program to defeat a world champion at Go. From Remarkable 2026: Presented at Vector Institute's annual conference (February 19-20, 2026), where 1,500+ researchers and industry leaders explored how AI research translates into real-world impact. This talk was part of day one's breakthrough research presentations addressing the shift from "What can AI answer?" to "What can AI do?" About Vector Institute: Vector drives excellence in Canada's AI knowledge creation and application, fostering economic growth through research, industry collaboration, and talent development. We're building Canada's leadership in responsible AI innovation. Join the Community: 🌐 https://vectorinstitute.ai/ 🐦 https://x.com/VectorInst 💼   / vector-institute   Subscribe for more cutting-edge AI research presentations, conference talks, and insights from leading researchers.