Category theory in machine learning: rewriting models for time series

In this project, we give a categorical perspective on the class of dynamical-system models in machine learning, where processes are iteratively applied over time. We develop an algebraic description of these models as a measure of compositionality: the properties of a model that are preserved under composition.

Iolo Jones, Jerry Swan, and Jeffrey Giansiracusa. Algebraic Dynamical Systems in Machine Learning. Applied Categorical Structures 32(4), 2024. https://doi.org/10.1007/s10485-023-09762-9