Asymptotic freeness of layerwise Jacobians caused by invariance of multilayer perceptron: the Haar orthogonal case

DOI10.1007/s00220-022-04441-7OpenAlexW4309232451MaRDI QIDQ2699717

Tomohiro Hayase, Benoit Collins

Publication date: 19 April 2023

Published in: Communications in Mathematical Physics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2103.13466

Mathematics Subject Classification ID

Free probability and free operator algebras (46L54) Limit theorems in probability theory (60Fxx) Probability theory on algebraic and topological structures (60Bxx)

Related Items (2)

Eigenvalue distribution of large random matrices arising in deep neural networks: Orthogonal case ⋮ The Law of Multiplication of Large Random Matrices Revisited

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Free probability and random matrices
• Integration with respect to the Haar measure on unitary, orthogonal and symplectic group
• Information geometry and its applications
• Limit laws for random matrices and free products
• Products of many large random matrices and gradients in deep neural networks
• Multiplication of free random variables and the \(S\)-transform: the case of vanishing mean
• Limit theorems for uniform distributions on spheres in high-dimensional euclidean spaces
• Universal statistics of Fisher information in deep neural networks: mean field approach^*
• Probability

This page was built for publication: Asymptotic freeness of layerwise Jacobians caused by invariance of multilayer perceptron: the Haar orthogonal case