AMOEBA-BASED NEUROCOMPUTING WITH CHAOTIC DYNAMICS.

This article describes amoeba-based neurocomputing as a deadlock-breaking form of parallel computing used to search for reasonable solutions. The contraction-relaxation rhythmic oscillations of this unicellular organism uses shape deformations as an integrated computational capacity. The article dis...

Descripción completa

Detalles Bibliográficos
Publicado en:Communications of the ACM Vol. 50; no. 9; pp. 69 - 73
Autores principales: Aono, Masashi, Hara, Masahiko, Aihara, Kazuyuki
Formato: Artículo
Publicado: Association for Computing Machinery Sep2007
Materias:
Acceso en línea:Ver este registro en EBSCOhost
Descripción
Sumario:This article describes amoeba-based neurocomputing as a deadlock-breaking form of parallel computing used to search for reasonable solutions. The contraction-relaxation rhythmic oscillations of this unicellular organism uses shape deformations as an integrated computational capacity. The article discusses experiments in which a neural network is modeled on the amoeba's photoavoidance-based shape deformation under optical feedback control. This system is the first non-silicon based implementation of chaotic neural computing. The capability of spontaneously escaping from equilibrium and stability is essential for our biologically inspired computing, and amoeba chaotic dynamics achieve that.