On the weight dynamics of learning networks

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Sharafi, Nahal, Martin, Christoph, Hallerberg, Sarah
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916231902134272
author Sharafi, Nahal
Martin, Christoph
Hallerberg, Sarah
author_facet Sharafi, Nahal
Martin, Christoph
Hallerberg, Sarah
contents Neural networks have become a widely adopted tool for tackling a variety of problems in machine learning and artificial intelligence. In this contribution we use the mathematical framework of local stability analysis to gain a deeper understanding of the learning dynamics of feed forward neural networks. Therefore, we derive equations for the tangent operator of the learning dynamics of three-layer networks learning regression tasks. The results are valid for an arbitrary numbers of nodes and arbitrary choices of activation functions. Applying the results to a network learning a regression task, we investigate numerically, how stability indicators relate to the final training-loss. Although the specific results vary with different choices of initial conditions and activation functions, we demonstrate that it is possible to predict the final training loss, by monitoring finite-time Lyapunov exponents or covariant Lyapunov vectors during the training process.
format Preprint
id arxiv_https___arxiv_org_abs_2405_00743
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle On the weight dynamics of learning networks
Sharafi, Nahal
Martin, Christoph
Hallerberg, Sarah
Machine Learning
Chaotic Dynamics
Neural networks have become a widely adopted tool for tackling a variety of problems in machine learning and artificial intelligence. In this contribution we use the mathematical framework of local stability analysis to gain a deeper understanding of the learning dynamics of feed forward neural networks. Therefore, we derive equations for the tangent operator of the learning dynamics of three-layer networks learning regression tasks. The results are valid for an arbitrary numbers of nodes and arbitrary choices of activation functions. Applying the results to a network learning a regression task, we investigate numerically, how stability indicators relate to the final training-loss. Although the specific results vary with different choices of initial conditions and activation functions, we demonstrate that it is possible to predict the final training loss, by monitoring finite-time Lyapunov exponents or covariant Lyapunov vectors during the training process.
title On the weight dynamics of learning networks
topic Machine Learning
Chaotic Dynamics
url https://arxiv.org/abs/2405.00743