Energy-based analog neural network framework
Over the past decade a body of work has emerged and shown the disruptive potential of neuromorphic systems across a broad range of studies, often combining novel machine learning models and nanotechnologies. Still, the scope of investigations often remains limited to simple problems since the proces...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Frontiers Media S.A.
2023-03-01
|
Series: | Frontiers in Computational Neuroscience |
Subjects: | |
Online Access: | https://www.frontiersin.org/articles/10.3389/fncom.2023.1114651/full |
_version_ | 1811160817343660032 |
---|---|
author | Mohamed Watfa Mohamed Watfa Alberto Garcia-Ortiz Gilles Sassatelli |
author_facet | Mohamed Watfa Mohamed Watfa Alberto Garcia-Ortiz Gilles Sassatelli |
author_sort | Mohamed Watfa |
collection | DOAJ |
description | Over the past decade a body of work has emerged and shown the disruptive potential of neuromorphic systems across a broad range of studies, often combining novel machine learning models and nanotechnologies. Still, the scope of investigations often remains limited to simple problems since the process of building, training, and evaluating mixed-signal neural models is slow and laborious. In this paper, we introduce an open-source framework, called EBANA, that provides a unified, modularized, and extensible infrastructure, similar to conventional machine learning pipelines, for building and validating analog neural networks (ANNs). It uses Python as interface language with a syntax similar to Keras, while hiding the complexity of the underlying analog simulations. It already includes the most common building blocks and maintains sufficient modularity and extensibility to easily incorporate new concepts, electrical, and technological models. These features make EBANA suitable for researchers and practitioners to experiment with different design topologies and explore the various tradeoffs that exist in the design space. We illustrate the framework capabilities by elaborating on the increasingly popular Energy-Based Models (EBMs), used in conjunction with the local Equilibrium Propagation (EP) training algorithm. Our experiments cover 3 datasets having up to 60,000 entries and explore network topologies generating circuits in excess of 1,000 electrical nodes that can be extensively benchmarked with ease and in reasonable time thanks to the native EBANA parallelization capability. |
first_indexed | 2024-04-10T06:04:00Z |
format | Article |
id | doaj.art-7eeed83e220d497b9d9704921e1c923d |
institution | Directory Open Access Journal |
issn | 1662-5188 |
language | English |
last_indexed | 2024-04-10T06:04:00Z |
publishDate | 2023-03-01 |
publisher | Frontiers Media S.A. |
record_format | Article |
series | Frontiers in Computational Neuroscience |
spelling | doaj.art-7eeed83e220d497b9d9704921e1c923d2023-03-03T05:15:17ZengFrontiers Media S.A.Frontiers in Computational Neuroscience1662-51882023-03-011710.3389/fncom.2023.11146511114651Energy-based analog neural network frameworkMohamed Watfa0Mohamed Watfa1Alberto Garcia-Ortiz2Gilles Sassatelli3LIRMM, University of Montpellier, CNRS, Montpellier, FranceITEM, University of Bremen, Bremen, GermanyITEM, University of Bremen, Bremen, GermanyLIRMM, University of Montpellier, CNRS, Montpellier, FranceOver the past decade a body of work has emerged and shown the disruptive potential of neuromorphic systems across a broad range of studies, often combining novel machine learning models and nanotechnologies. Still, the scope of investigations often remains limited to simple problems since the process of building, training, and evaluating mixed-signal neural models is slow and laborious. In this paper, we introduce an open-source framework, called EBANA, that provides a unified, modularized, and extensible infrastructure, similar to conventional machine learning pipelines, for building and validating analog neural networks (ANNs). It uses Python as interface language with a syntax similar to Keras, while hiding the complexity of the underlying analog simulations. It already includes the most common building blocks and maintains sufficient modularity and extensibility to easily incorporate new concepts, electrical, and technological models. These features make EBANA suitable for researchers and practitioners to experiment with different design topologies and explore the various tradeoffs that exist in the design space. We illustrate the framework capabilities by elaborating on the increasingly popular Energy-Based Models (EBMs), used in conjunction with the local Equilibrium Propagation (EP) training algorithm. Our experiments cover 3 datasets having up to 60,000 entries and explore network topologies generating circuits in excess of 1,000 electrical nodes that can be extensively benchmarked with ease and in reasonable time thanks to the native EBANA parallelization capability.https://www.frontiersin.org/articles/10.3389/fncom.2023.1114651/fullneural networksenergy-based modelsequilibrium propagationframeworkanalogmixed-signal |
spellingShingle | Mohamed Watfa Mohamed Watfa Alberto Garcia-Ortiz Gilles Sassatelli Energy-based analog neural network framework Frontiers in Computational Neuroscience neural networks energy-based models equilibrium propagation framework analog mixed-signal |
title | Energy-based analog neural network framework |
title_full | Energy-based analog neural network framework |
title_fullStr | Energy-based analog neural network framework |
title_full_unstemmed | Energy-based analog neural network framework |
title_short | Energy-based analog neural network framework |
title_sort | energy based analog neural network framework |
topic | neural networks energy-based models equilibrium propagation framework analog mixed-signal |
url | https://www.frontiersin.org/articles/10.3389/fncom.2023.1114651/full |
work_keys_str_mv | AT mohamedwatfa energybasedanalogneuralnetworkframework AT mohamedwatfa energybasedanalogneuralnetworkframework AT albertogarciaortiz energybasedanalogneuralnetworkframework AT gillessassatelli energybasedanalogneuralnetworkframework |