Memristors Empower Spiking Neurons With Stochasticity

Handle URI:
http://hdl.handle.net/10754/575739
Title:
Memristors Empower Spiking Neurons With Stochasticity
Authors:
Al-Shedivat, Maruan ( 0000-0001-9037-1005 ) ; Naous, Rawan ( 0000-0001-6129-7926 ) ; Cauwenberghs, Gert; Salama, Khaled N. ( 0000-0001-7742-1282 )
Abstract:
Recent theoretical studies have shown that probabilistic spiking can be interpreted as learning and inference in cortical microcircuits. This interpretation creates new opportunities for building neuromorphic systems driven by probabilistic learning algorithms. However, such systems must have two crucial features: 1) the neurons should follow a specific behavioral model, and 2) stochastic spiking should be implemented efficiently for it to be scalable. This paper proposes a memristor-based stochastically spiking neuron that fulfills these requirements. First, the analytical model of the memristor is enhanced so it can capture the behavioral stochasticity consistent with experimentally observed phenomena. The switching behavior of the memristor model is demonstrated to be akin to the firing of the stochastic spike response neuron model, the primary building block for probabilistic algorithms in spiking neural networks. Furthermore, the paper proposes a neural soma circuit that utilizes the intrinsic nondeterminism of memristive switching for efficient spike generation. The simulations and analysis of the behavior of a single stochastic neuron and a winner-take-all network built of such neurons and trained on handwritten digits confirm that the circuit can be used for building probabilistic sampling and pattern adaptation machinery in spiking networks. The findings constitute an important step towards scalable and efficient probabilistic neuromorphic platforms. © 2011 IEEE.
KAUST Department:
Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division; Computer Science Program; Electrical Engineering Program; Sensors Lab
Publisher:
Institute of Electrical and Electronics Engineers (IEEE)
Journal:
IEEE Journal on Emerging and Selected Topics in Circuits and Systems
Issue Date:
Jun-2015
DOI:
10.1109/JETCAS.2015.2435512
Type:
Article
ISSN:
2156-3357; 2156-3365
Appears in Collections:
Articles; Computer Science Program; Computer Science Program; Electrical Engineering Program; Electrical Engineering Program; Sensors Lab; Sensors Lab; Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division

Full metadata record

DC FieldValue Language
dc.contributor.authorAl-Shedivat, Maruanen
dc.contributor.authorNaous, Rawanen
dc.contributor.authorCauwenberghs, Gerten
dc.contributor.authorSalama, Khaled N.en
dc.date.accessioned2015-08-24T08:36:56Zen
dc.date.available2015-08-24T08:36:56Zen
dc.date.issued2015-06en
dc.identifier.issn2156-3357en
dc.identifier.issn2156-3365en
dc.identifier.doi10.1109/JETCAS.2015.2435512en
dc.identifier.urihttp://hdl.handle.net/10754/575739en
dc.description.abstractRecent theoretical studies have shown that probabilistic spiking can be interpreted as learning and inference in cortical microcircuits. This interpretation creates new opportunities for building neuromorphic systems driven by probabilistic learning algorithms. However, such systems must have two crucial features: 1) the neurons should follow a specific behavioral model, and 2) stochastic spiking should be implemented efficiently for it to be scalable. This paper proposes a memristor-based stochastically spiking neuron that fulfills these requirements. First, the analytical model of the memristor is enhanced so it can capture the behavioral stochasticity consistent with experimentally observed phenomena. The switching behavior of the memristor model is demonstrated to be akin to the firing of the stochastic spike response neuron model, the primary building block for probabilistic algorithms in spiking neural networks. Furthermore, the paper proposes a neural soma circuit that utilizes the intrinsic nondeterminism of memristive switching for efficient spike generation. The simulations and analysis of the behavior of a single stochastic neuron and a winner-take-all network built of such neurons and trained on handwritten digits confirm that the circuit can be used for building probabilistic sampling and pattern adaptation machinery in spiking networks. The findings constitute an important step towards scalable and efficient probabilistic neuromorphic platforms. © 2011 IEEE.en
dc.publisherInstitute of Electrical and Electronics Engineers (IEEE)en
dc.subjectNeuromorphic systemsen
dc.subjectprobabilistic inferenceen
dc.subjectprobabilistic learningen
dc.subjectspiking neuronsen
dc.subjectstochastic computingen
dc.subjectstochastic memristorsen
dc.subjectwinner-take-allen
dc.titleMemristors Empower Spiking Neurons With Stochasticityen
dc.typeArticleen
dc.contributor.departmentComputer, Electrical and Mathematical Sciences and Engineering (CEMSE) Divisionen
dc.contributor.departmentComputer Science Programen
dc.contributor.departmentElectrical Engineering Programen
dc.contributor.departmentSensors Laben
dc.identifier.journalIEEE Journal on Emerging and Selected Topics in Circuits and Systemsen
dc.contributor.institutionInstitute for Neural Computation and the Department of Bioengineering, University of California, San Diego, La Jollen
kaust.authorAl-Shedivat, Maruanen
kaust.authorSalama, Khaled N.en
kaust.authorNaous, Rawanen
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.