Asymptotic description of stochastic neural networks. I: Existence of a large deviation principle

DOI10.1016/J.CRMA.2014.08.018MaRDI QIDQ467694zbMATH OpenOpenAlexFDO

Authors Olivier Faugeras, James Maclaurin

Publication date 4 November 2014

Published in Comptes Rendus. Mathématique. Académie des Sciences, Paris (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1407.2457

zbMATH Keywords

asymptotic behaviourlarge deviationsrandom variablesrate function

Mathematics Subject Classification ID

Large deviations (60F10) Interacting random processes; statistical mechanics type models; percolation theory (60K35) Neural networks for/in biological studies, artificial life and related topics (92B20)

Abstract: We study the asymptotic law of a network of interacting neurons when the number of neurons becomes infinite. The dynamics of the neurons is described by a set of stochastic differential equations in discrete time. The neurons interact through the synaptic weights which are Gaussian correlated random variables. We describe the asymptotic law of the network when the number of neurons goes to infinity. Unlike previous works which made the biologically unrealistic assumption that the weights were i.i.d. random variables, we assume that they are correlated. We introduce the process-level empirical measure of the trajectories of the solutions to the equations of the finite network of neurons and the averaged law (with respect to the synaptic weights) of the trajectories of the solutions to the equations of the network of neurons. The result is that the image law through the empirical measure satisfies a large deviation principle with a good rate function. We provide an analytical expression of this rate function in terms of the spectral representation of certain Gaussian processes.

Review text

It is considered a completely connected network of neurons in which the synaptic weights are Gaussian-correlated random variables. The aim of the article is to investigate the asymptotic behaviour and large deviations of the network when the number of neurons goes to infinity. In the case where the synaptic weights are i.i.d. random variables, with Gaussian zero mean, this problem has been investigated by \textit{H. Sompolinsky} et al. in [``Chaos in random neural networks, Phys. Rev. Lett. 61, 259--262 (1988; \url{doi:10.1103/PhysRevLett.61.259})]. The present article presents a mathematical framework, including definitions, model and preliminaries. The main result states that: ``\( \Pi^n\) is governed by a large deviation principle with a good rate function \(H\), is just announced and not proved. The authors claim that the proof can not be reproduced here because it is too long, the main steps are indicated.NEWLINENEWLINEFor details and proofs, the reader should consult the complete paper by the authors [``A large deviation principle for networks of rate neurons with correlated synaptic weights, Preprint, \url{arXiv:1302.1029}].

Recommendations

• Asymptotic description of stochastic neural networks. II: Characterization of the limit law
• Large deviations for randomly connected neural networks. II: State-dependent interactions
• Large deviations and mean-field theory for asymmetric random recurrent neural networks
• Large deviations, dynamics and phase transitions in large stochastic and disordered neural networks
• Large deviations for randomly connected neural networks. I: Spatially extended systems

Cites work

• A law of large numbers for random walks in random environment
• Asymptotic description of stochastic neural networks. I: Existence of a large deviation principle
• Asymptotic description of stochastic neural networks. II: Characterization of the limit law
• Large deviations and mean-field theory for asymmetric random recurrent neural networks
• Microcanonical distributions for lattice gases
• The point of view of the particle on the law of large numbers for random walks in a mixing random environment

Cited in

(11)

• Mean-field limit of generalized Hawkes processes
• Large deviations and mean-field theory for asymmetric random recurrent neural networks
• Besicovitch almost automorphic stochastic processes in distribution and an application to Clifford-valued stochastic neural networks
• Large deviations for nonlocal stochastic neural fields
• Quenched large deviations for interacting diffusions in random media
• A formalism for evaluating analytically the cross-correlation structure of a firing-rate network model
• A large deviation principle and an expression of the rate function for a discrete stationary Gaussian process
• Asymptotic description of stochastic neural networks. I: Existence of a large deviation principle
• Asymptotic description of stochastic neural networks. II: Characterization of the limit law
• Large deviations for randomly connected neural networks. I: Spatially extended systems
• Large deviations for randomly connected neural networks. II: State-dependent interactions

This page was built for publication: Asymptotic description of stochastic neural networks. I: Existence of a large deviation principle

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q467694)