Fixed points of competitive threshold-linear networks

Carina Pamela Curto, Jesse Geneson, Katherine Morrison

Research output: Contribution to journalLetter

2 Citations (Scopus)

Abstract

Threshold-linear networks (TLNs) are models of neural networks that consist of simple, perceptron-like neurons and exhibit nonlinear dynamics determined by the network's connectivity. The fixed points of a TLN, including both stable and unstable equilibria, play a critical role in shaping its emergent dynamics. In this work,we provide two novel characterizations for the set of fixed points of a competitive TLN: the first is in terms of a simple sign condition, while the second relies on the concept of domination. We apply these results to a special family of TLNs, called combinatorial threshold-linear networks (CTLNs),whose connectivity matrices are defined from directed graphs. This leads us to prove a series of graph rules that enable one to determine fixed points of a CTLN by analyzing the underlying graph. In addition, we study larger networks composed of smaller building block subnetworks and prove several theorems relating the fixed points of the full network to those of its components. Our results provide the foundation for a kind of graphical calculus to infer features of the dynamics from a network's connectivity.

Original languageEnglish (US)
Pages (from-to)94-155
Number of pages62
JournalNeural Computation
Volume31
Issue number1
DOIs
StatePublished - Jan 1 2019

Fingerprint

Neural Networks (Computer)
Nonlinear Dynamics
Calculi
Linear Models
Neurons
Fixed Point

All Science Journal Classification (ASJC) codes

  • Arts and Humanities (miscellaneous)
  • Cognitive Neuroscience

Cite this

Curto, Carina Pamela ; Geneson, Jesse ; Morrison, Katherine. / Fixed points of competitive threshold-linear networks. In: Neural Computation. 2019 ; Vol. 31, No. 1. pp. 94-155.
@article{bc4d1d478e414ba5bae46e9c9b95e094,
title = "Fixed points of competitive threshold-linear networks",
abstract = "Threshold-linear networks (TLNs) are models of neural networks that consist of simple, perceptron-like neurons and exhibit nonlinear dynamics determined by the network's connectivity. The fixed points of a TLN, including both stable and unstable equilibria, play a critical role in shaping its emergent dynamics. In this work,we provide two novel characterizations for the set of fixed points of a competitive TLN: the first is in terms of a simple sign condition, while the second relies on the concept of domination. We apply these results to a special family of TLNs, called combinatorial threshold-linear networks (CTLNs),whose connectivity matrices are defined from directed graphs. This leads us to prove a series of graph rules that enable one to determine fixed points of a CTLN by analyzing the underlying graph. In addition, we study larger networks composed of smaller building block subnetworks and prove several theorems relating the fixed points of the full network to those of its components. Our results provide the foundation for a kind of graphical calculus to infer features of the dynamics from a network's connectivity.",
author = "Curto, {Carina Pamela} and Jesse Geneson and Katherine Morrison",
year = "2019",
month = "1",
day = "1",
doi = "10.1162/neco_a_01151",
language = "English (US)",
volume = "31",
pages = "94--155",
journal = "Neural Computation",
issn = "0899-7667",
publisher = "MIT Press Journals",
number = "1",

}

Fixed points of competitive threshold-linear networks. / Curto, Carina Pamela; Geneson, Jesse; Morrison, Katherine.

In: Neural Computation, Vol. 31, No. 1, 01.01.2019, p. 94-155.

Research output: Contribution to journalLetter

TY - JOUR

T1 - Fixed points of competitive threshold-linear networks

AU - Curto, Carina Pamela

AU - Geneson, Jesse

AU - Morrison, Katherine

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Threshold-linear networks (TLNs) are models of neural networks that consist of simple, perceptron-like neurons and exhibit nonlinear dynamics determined by the network's connectivity. The fixed points of a TLN, including both stable and unstable equilibria, play a critical role in shaping its emergent dynamics. In this work,we provide two novel characterizations for the set of fixed points of a competitive TLN: the first is in terms of a simple sign condition, while the second relies on the concept of domination. We apply these results to a special family of TLNs, called combinatorial threshold-linear networks (CTLNs),whose connectivity matrices are defined from directed graphs. This leads us to prove a series of graph rules that enable one to determine fixed points of a CTLN by analyzing the underlying graph. In addition, we study larger networks composed of smaller building block subnetworks and prove several theorems relating the fixed points of the full network to those of its components. Our results provide the foundation for a kind of graphical calculus to infer features of the dynamics from a network's connectivity.

AB - Threshold-linear networks (TLNs) are models of neural networks that consist of simple, perceptron-like neurons and exhibit nonlinear dynamics determined by the network's connectivity. The fixed points of a TLN, including both stable and unstable equilibria, play a critical role in shaping its emergent dynamics. In this work,we provide two novel characterizations for the set of fixed points of a competitive TLN: the first is in terms of a simple sign condition, while the second relies on the concept of domination. We apply these results to a special family of TLNs, called combinatorial threshold-linear networks (CTLNs),whose connectivity matrices are defined from directed graphs. This leads us to prove a series of graph rules that enable one to determine fixed points of a CTLN by analyzing the underlying graph. In addition, we study larger networks composed of smaller building block subnetworks and prove several theorems relating the fixed points of the full network to those of its components. Our results provide the foundation for a kind of graphical calculus to infer features of the dynamics from a network's connectivity.

UR - http://www.scopus.com/inward/record.url?scp=85059485880&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85059485880&partnerID=8YFLogxK

U2 - 10.1162/neco_a_01151

DO - 10.1162/neco_a_01151

M3 - Letter

C2 - 30462583

AN - SCOPUS:85059485880

VL - 31

SP - 94

EP - 155

JO - Neural Computation

JF - Neural Computation

SN - 0899-7667

IS - 1

ER -