NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET

Many currently available hardware implementations of the unsupervised self-organizing feature map (SOFM) algorithm utilize complementary metal–oxide–semiconductor (CMOS)-only circuits that often compromise key behaviors of the SOFM algorithm due to complexity. We propose a neur...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Siddharth Barve, Joshua Mayersky, Andrew J. Ford, Alexander Jones, Bayley King, Aaron Ruen, Rashmi Jha
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Ferroelectric field-effect transistor neuromorphic resistive random access memory (RRAM) self-organizing feature map (SOFM) unsupervised Computer engineering. Computer hardware TK7885-7895
Acceso en línea:	https://doaj.org/article/66f3762c4a0945d7a60e92ccad61c13f
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:66f3762c4a0945d7a60e92ccad61c13f
record_format	dspace
spelling	oai:doaj.org-article:66f3762c4a0945d7a60e92ccad61c13f2021-11-03T23:00:13ZNeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET2329-923110.1109/JXCDC.2021.3119489https://doaj.org/article/66f3762c4a0945d7a60e92ccad61c13f2021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9568851/https://doaj.org/toc/2329-9231Many currently available hardware implementations of the unsupervised self-organizing feature map (SOFM) algorithm utilize complementary metal–oxide–semiconductor (CMOS)-only circuits that often compromise key behaviors of the SOFM algorithm due to complexity. We propose a neuromorphic architecture harnessing the unique properties of ferroelectric field-effect transistors (FeFETs) and gated-resistive random access memory (RRAM) for in-memory computing to implement the SOFM algorithm. The FeFET-based synapse, organized in a novel circuit, is able to compute the input-weight Euclidean error in memory via the saturation drain current. The self-decaying states of the gated-RRAM allow for a self-decaying neighborhood and learning rate implementation to allow for convergence and lifelong learning. This novel architecture is able to successfully cluster benchmarks (RGB colors and MNIST handwritten digits) and real-life datasets, such as COVID-19 patient chest X-rays completely unsupervised. The architecture also demonstrates a significant amount of robustness to device variability and damaged neurons. In addition, the proposed architecture is completely parallelized and provides a power-efficient platform for implementing the SOFM algorithm.Siddharth BarveJoshua MayerskyAndrew J. FordAlexander JonesBayley KingAaron RuenRashmi JhaIEEEarticleFerroelectric field-effect transistorneuromorphicresistive random access memory (RRAM)self-organizing feature map (SOFM)unsupervisedComputer engineering. Computer hardwareTK7885-7895ENIEEE Journal on Exploratory Solid-State Computational Devices and Circuits, Vol 7, Iss 2, Pp 97-105 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Ferroelectric field-effect transistor neuromorphic resistive random access memory (RRAM) self-organizing feature map (SOFM) unsupervised Computer engineering. Computer hardware TK7885-7895
spellingShingle	Ferroelectric field-effect transistor neuromorphic resistive random access memory (RRAM) self-organizing feature map (SOFM) unsupervised Computer engineering. Computer hardware TK7885-7895 Siddharth Barve Joshua Mayersky Andrew J. Ford Alexander Jones Bayley King Aaron Ruen Rashmi Jha NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
description	Many currently available hardware implementations of the unsupervised self-organizing feature map (SOFM) algorithm utilize complementary metal–oxide–semiconductor (CMOS)-only circuits that often compromise key behaviors of the SOFM algorithm due to complexity. We propose a neuromorphic architecture harnessing the unique properties of ferroelectric field-effect transistors (FeFETs) and gated-resistive random access memory (RRAM) for in-memory computing to implement the SOFM algorithm. The FeFET-based synapse, organized in a novel circuit, is able to compute the input-weight Euclidean error in memory via the saturation drain current. The self-decaying states of the gated-RRAM allow for a self-decaying neighborhood and learning rate implementation to allow for convergence and lifelong learning. This novel architecture is able to successfully cluster benchmarks (RGB colors and MNIST handwritten digits) and real-life datasets, such as COVID-19 patient chest X-rays completely unsupervised. The architecture also demonstrates a significant amount of robustness to device variability and damaged neurons. In addition, the proposed architecture is completely parallelized and provides a power-efficient platform for implementing the SOFM algorithm.
format	article
author	Siddharth Barve Joshua Mayersky Andrew J. Ford Alexander Jones Bayley King Aaron Ruen Rashmi Jha
author_facet	Siddharth Barve Joshua Mayersky Andrew J. Ford Alexander Jones Bayley King Aaron Ruen Rashmi Jha
author_sort	Siddharth Barve
title	NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
title_short	NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
title_full	NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
title_fullStr	NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
title_full_unstemmed	NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET
title_sort	neurosofm: a neuromorphic self-organizing feature map heterogeneously integrating rram and fefet
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/66f3762c4a0945d7a60e92ccad61c13f
work_keys_str_mv	AT siddharthbarve neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT joshuamayersky neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT andrewjford neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT alexanderjones neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT bayleyking neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT aaronruen neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet AT rashmijha neurosofmaneuromorphicselforganizingfeaturemapheterogeneouslyintegratingrramandfefet
_version_	1718445334141599744

NeuroSOFM: A Neuromorphic Self-Organizing Feature Map Heterogeneously Integrating RRAM and FeFET

Ejemplares similares