Paper
1 April 1991 Two-dimensional dynamic neural network optical system with simplest types of large-scale interactions
Author Affiliations +
Proceedings Volume 1402, USSR-CSFR Joint Seminar on Nonlinear Optics in Control, Diagnostics, and Modeling of Biophysical Processes; (1991) https://doi.org/10.1117/12.47489
Event: USSR-CSFR Joint Seminar on Nonlinear Optics in Control, Diagnostics and Modeling, 1990, Koshice-Prague, Czechoslovakia
Abstract
Modern nonlinear and coherent optics provides a great variety of possibilities to create many types of spatially distributed neural network optical systems and uses them to simulate the collective behavior of dynamics systems with nonlocal interactions (e.g., analogs of advanced neural networks).' We can point out the following basic properties of these systems: — intrinsic N-shape nonlinearity (bistable or multistable optical response); — existence of the local excitation transfer mechanism (e.g., diffusion); — possible implementation of large scale (nonlocal) interactions; All these requirements can be fulfilled in nonlinear 2-D feedback optical systems: in various nonlinear resonators and interferometers.2 Complexity and variety of nonlinear dynamic modes (different selfoscillation and dissipative structures, optical turbulence) observed in neural network optical systems with the simplest nonlocal interconnectors promote the search for principal most important models of large scale interactions and corresponding nonlinear dynamic structures. One of these models is one-dimensional rotational instability occurring in a neural network system with rotated field.3 Rotating waves observed in this system have been analyzed in detail.4 Theoretical and experimental research of 2-D models of optical neural networks appears to be most interesting. Implementing any type of "pure" nonlocal interconnectors in an optical experiment is a rather complicated task. Optical aberrations and some other factors result in the fact that additional "spurious" interactions affecting dynamic structures being formed in the system seem to be superimposed on a given type of nonlocal interconnectors. The only possibility to examine multiple nonlinear structures corresponding to a certain type of large scale interactions appears to be a comparative analysis of optical and numerical experiments in which the interconnect topologies can be defined in a "pure" form. 2-D selfoscillation structures observed in a neural network with 2-D optical feedback and nonlocal interconnectors have been studied in this paper. The simplest types of topological transforms, i.e., lateral inversion and rotation, are realized in optical and numerical experiments.
© (1991) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Mikhail A. Vorontsov, N. I. Zheleznykh, and Andrey V. Larichev "Two-dimensional dynamic neural network optical system with simplest types of large-scale interactions", Proc. SPIE 1402, USSR-CSFR Joint Seminar on Nonlinear Optics in Control, Diagnostics, and Modeling of Biophysical Processes, (1 April 1991); https://doi.org/10.1117/12.47489
Lens.org Logo
CITATIONS
Cited by 1 scholarly publication.
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Diffusion

Process modeling

Neural networks

Nonlinear optics

Feedback loops

Diagnostics

Complex systems

Back to Top