Exact and Approximate Symmetries in Machine Learning Models

Explore the concept of exact and approximate symmetries in machine learning models in this 58-minute conference talk by Soledad Villar from John Hopkins University. Delve into the intricate relationship between symmetries and machine learning, examining how these principles influence model design and performance. Gain insights into the mathematical foundations that underpin symmetry in AI systems, and discover how researchers are leveraging both exact and approximate symmetries to enhance model efficiency and generalization. Learn about cutting-edge approaches that incorporate symmetry considerations into various machine learning architectures, and understand the potential implications for fields ranging from computer vision to natural language processing. Engage with complex ideas presented by a leading expert in the field as part of the AI and Pure Mathematics Conference hosted by the University of Miami.

Soledad Villar, John Hopkins University: Exact and approximate symmetries in machine learning models