ARTIFICIAL NEURAL NETWORKS: FROM MATHEMATICAL MODELS TO BIOLOGICALLY INSPIRED SELF-ORGANIZING SYSTEMS

This paper provides a comprehensive analysis of the evolutionary development of artificial neural networks (ANNs) through the lens of three key generations: from simple perceptrons to modern spiking neural networks (SNNs) and prospective biophysical models. Particular attention is paid to a critical comparison of artificial systems with their biological prototypes, identifying the fundamental limitations of existing approaches, and justifying the need for a new paradigmatic direction - self-organizing networks of uniform elements (SNUE). The proposed SNUE concept integrates key principles of biological neuroplasticity with the requirements of computational efficiency, offering an innovative framework for the development of the next generation of neuromorphic systems. The paper provides a detailed analysis of the theoretical foundations, potential architectural solutions, and promising directions for the practical implementation of this approach.