Tensor Species: The Theory and Practice of Neural Networks

Explore a 59-minute Topos Institute Colloquium talk that bridges category theory and machine learning, focusing on how Joyal's theory of species can be applied to understand and describe data types in modern neural networks. Delve into the challenges of finding practical abstractions for neural network design innovation while maintaining a performance-oriented perspective. Learn how theoretical frameworks from category theory can provide concrete insights into the structure and behavior of contemporary neural network architectures.