Designing a neural network via gradient flow

  • Authors:
  • Nicoletta Del Buono

  • Affiliations:
  • Dipartimento Interuniversitario di Matematica Università degli Studi di Bari via E. Orabona 4, I-70125 Bari, ITALY

  • Venue:
  • Second international workshop on Intelligent systems design and application
  • Year:
  • 2002

Quantified Score

Hi-index 0.01

Visualization

Abstract

The problem of designing a neural network possessing prescribed stable equilibria is a fundamental task for network designers. This problem can be viewed as an inverse eigenvalue problem and solved by using a gradient flow approach. Working with discretization of continuous-time systems could produce new algorithms to solve the problem and could allow the possibility of constructing adaptive schemes faster than algebraic approaches.