Neural Networks as Geometric Chaotic Maps

arXiv1912.05081MaRDI QIDQ6330873

Author name not available (Why is that?)

Publication date: 10 December 2019

Abstract: The use of artificial neural networks as models of chaotic dynamics has been rapidly expanding. Still, a theoretical understanding of how neural networks learn chaos is lacking. Here, we employ a geometric perspective to show that neural networks can efficiently model chaotic dynamics by becoming structurally chaotic themselves. We first confirm neural network's efficiency in emulating chaos by showing that a parsimonious neural network trained only on few data points can reconstruct strange attractors, extrapolate outside training data boundaries, and accurately predict local divergence rates. We then posit that the trained network's map comprises sequential geometric stretching, rotation, and compression operations. These geometric operations indicate topological mixing and chaos, explaining why neural networks are naturally suitable to emulate chaotic dynamics.

Has companion code repository: https://github.com/dante831/NN_Lorenz

This page was built for publication: Neural Networks as Geometric Chaotic Maps

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6330873)