Competition aided with finite-time neural network

Shuai Li, Long Jin

Research output: Chapter in book / Conference proceedingChapter in an edited book (as author)Academic researchpeer-review

Abstract

In this chapter, a class of recurrent neural networks to solve quadratic programming problems are presented and further extended to competition generation. Different from most existing recurrent neural networks for solving quadratic programming problems, the proposed neural network model converges in finite time and the activation function is not required to be a hard-limiting function for finite convergence time. The stability, finite-time convergence property and the optimality of the proposed neural network for solving the original quadratic programming problem are proven in theory. Extensive simulations are performed to evaluate the performance of the neural network with different parameters. In addition, the proposed neural network is applied to solving the k-winner-take-all (k-WTA) problem. Both theoretical analysis and numerical simulations validate the effectiveness of our method for solving the k-WTA problem.

Original languageEnglish
Title of host publicationSpringerBriefs in Applied Sciences and Technology
PublisherSpringer-Verlag
Pages25-55
Number of pages31
Edition9789811049460
DOIs
Publication statusPublished - 1 Jan 2018

Publication series

NameSpringerBriefs in Applied Sciences and Technology
Number9789811049460
ISSN (Print)2191-530X
ISSN (Electronic)2191-5318

Keywords

  • Finite-time convergence
  • Global stability
  • Numerical simulations
  • Quadratic programming
  • Recurrent neural networks
  • Winner-take-all competition

ASJC Scopus subject areas

  • Biotechnology
  • Chemical Engineering(all)
  • Mathematics(all)
  • Materials Science(all)
  • Energy Engineering and Power Technology
  • Engineering(all)

Cite this