Editing Victor Eliashberg

Title: Victor Eliashberg

Main Research Question: How can "dynamical" neural networks function as universal programmable "symbolic" machines?

Methodology: The study uses the concept of E-machine, a hypothetical brain-like dynamically reconfigurable associative learning system. This concept is applied to various applications such as context-sensitive associative memory, context-dependent pattern classification, context-dependent motor control, imitation, simulation of complex "informal" environments, and natural language.

Results: The research suggests that "dynamical" neural networks can function as universal programmable "symbolic" machines. It also proposes that a universal learning machine can dynamically reconfigure its software depending on a combinatorial number of possible contexts.

Implications: This study has significant implications for the understanding of human brain's remarkable information processing characteristics. It sheds light on the brain's universality as a learning system and its ability to dynamically change its behavior depending on a combinatorial number of different contexts. The concept of E-machine provides a neurobiologically consistent formalization of these processes, connecting symbolic and dynamical computational mechanisms in a way that is relative ly short and formal.

Link to Article: https://arxiv.org/abs/0403031v1
Authors: 
arXiv ID: 0403031v1

[[Category:Computer Science]]
[[Category:As]]
[[Category:Can]]
[[Category:Dynamical]]
[[Category:Universal]]
[[Category:Symbolic]]