This web content explores the application of Noether’s principle, a fundamental concept in physics that establishes a relationship between invariances of the action and conserved quantities in a physical system. The discussion extends this principle to machine learning processes, investigating whether similar conserved quantities can be found in discrete processes such as deep learning models. The content delves into the mathematical principles behind Noether’s theorem, offering a detailed toy example to illustrate how invariances under certain transformations lead to conserved quantities. The application of Noether’s principle in machine learning poses intriguing questions about the potential implications for understanding interactions between components in complex models.
https://cgad.ski/blog/where-is-noethers-principle-in-machine-learning.html