Machine Learning Street Talk (MLST) #60 Geometric Deep Learning Blueprint (Special Edition)
64 snips
Sep 19, 2021 
Joining the discussion are Petar Veličković from DeepMind, renowned for his work on graph neural networks, Taco Cohen from Qualcomm AI Research, specializing in geometric deep learning, and Joan Bruna, an influential figure in data science at NYU. They delve into geometric deep learning, exploring its foundations in symmetry and invariance. The conversation highlights innovative mathematical frameworks, the unification of geometries, and their implications in AI. Insights on dimensionality, algorithmic reasoning, and historical perspectives on geometry further enrich this engaging dialogue.
AI Snips
Chapters
Books
Transcript
Episode notes
Geometric Deep Learning and the Curse of Dimensionality
- High-dimensional learning requires strong assumptions about function regularities, like smoothness.
- Geometric priors, by leveraging low-dimensional structure, reduce overfitting risk.
Geometric Deep Learning Blueprint
- The Geometric Deep Learning blueprint provides a framework for understanding and building deep learning architectures.
- It emphasizes symmetry, scale separation, and geometric stability as core principles.
Geometric Deep Learning in Drug Discovery
- Michael Bronstein's team used Geometric Deep Learning to predict anti-cancer drug properties in molecules.
- This framework addresses the challenge of traditional ML with network-structured data.
