Particle/Astro-Machine Learning Seminar

Abstract:

We propose a theoretical understanding of neural networks in terms of Wilsonian effective field theory. The correspondence relies on the fact that many asymptotic neural networks are drawn from Gaussian processes, the analog of non-interacting field theories. Moving away from the asymptotic limit yields a non-Gaussian process and corresponds to turning on particle in- teractions, allowing for the computation of correlation functions of neural network outputs with Feynman diagrams.