From Learning Differential Operators to Learning Algorithms

Explore the intersection of differential operators and algorithmic learning in this mathematical sciences lecture delivered by Professor Houman Owhadi from Caltech at the Isaac Newton Institute. Delve into advanced mathematical concepts that bridge the gap between traditional differential operator theory and modern machine learning algorithms. Examine how principles from differential equations can inform and enhance algorithmic approaches to learning problems. Discover the theoretical foundations that connect these seemingly disparate mathematical domains and understand their practical applications in computational mathematics. Learn about cutting-edge research methodologies that leverage differential operator frameworks to develop more sophisticated learning algorithms. Gain insights into how classical mathematical analysis techniques can be adapted and applied to contemporary machine learning challenges. This presentation is part of the "Representing, calibrating & leveraging prediction uncertainty from statistics to machine learning" research programme, offering a unique perspective on the mathematical underpinnings of uncertainty quantification in learning systems.