D-127 | Memristive Hardware Neural Networks

D-127 | Memristive Hardware Neural Networks 150 150 SAN 2024 Annual Meeting

Theoretical and Computational Neuroscience
Author: Leandro Ezequiel Fernandez | Email: leandrofernandez671@gmail.com


Leandro Ezequiel Fernandez1°2°, Marcelo Rozenberg, Gabriel Bernardo Mindlin1°2°,  , , , ,  , ,

Dpto. de Física, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires
Instituto de Física Interdiciplinaria y Aplicada, INFINA-CONICET, Buenos Aires, Argentina
Université Paris-Saclay, CNRS Laboratoire de Physique des Solides, 91405, Orsay, France

In the last years, many mathematical models have been proposed to emulate the behavior of neurons. Electronic circuits which integrate analogically those equations, or display dynamics qualitatively similar to neurons, are known as electronic neurons. The advantage of performing an analogic integration of the equations, as opposed to integrating them by software running on a computer, emerges if the computation requires or is designed to interact on real time with a living being. In that case, an analog integration avoids any delays caused by an operative system controlling data acquisition cards, or the running of the codes themselves. In this work we built on previous efforts that lead to the construction of an electronic neuron of minimal complexity in its design, and yet was capable of displaying a diversity of excitable dynamics. We now show how to couple the units in a realistic way and build the first elements of a circuit capable of displaying acoustically selective responses.

Masterfully Handcrafted for Awesomeness

WE DO MOVE

YOUR WORLD

Greatives – Design, Marketing, Sales

Working Hours : 09:00 – 19:00
Address : 44 Oxford Street, London, UK 22004
Phone : +380 22 333 555