A Memristor-Based Optimization Framework for Artificial Intelligence Applications

Sijia Liu, Yanzhi Wang, Makan Fardad, Pramod Kumar Varshney

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Memristors have recently received significant attention as device-level components for building a novel generation of computing systems. These devices have many promising features, such as non-volatility, low power consumption, high density, and excellent scalability. The ability to control and modify biasing voltages at memristor terminals make them promising candidates to efficiently perform matrix-vector multiplications and solve systems of linear equations. In this article, we discuss how networks of memristors arranged in crossbar arrays can be used for efficiently solving optimization and machine learning problems. We introduce a new memristor-based optimization framework that combines the computational merits of memristor crossbars with the advantages of an operator splitting method, the alternating direction method of multipliers (ADMM). Here, ADMM helps in splitting a complex optimization problem into subproblems that involve the solution of systems of linear equations. The strength of this framework is shown by applying it to linear programming, quadratic programming, and sparse optimization. In addition to ADMM, implementation of a customized power iteration method for eigenvalue/eigenvector computation using memristor crossbars is discussed. The memristor-based power iteration method can further be applied to principal component analysis. The use of memristor crossbars yields a significant speed-up in computation, and thus, we believe, has the potential to advance optimization and machine learning research in artificial intelligence.

Original languageEnglish (US)
Article number8288635
Pages (from-to)29-44
Number of pages16
JournalIEEE Circuits and Systems Magazine
Volume18
Issue number1
DOIs
StatePublished - Jan 1 2018

Fingerprint

Memristors
Artificial intelligence
Linear equations
Learning systems
Quadratic programming
Eigenvalues and eigenfunctions
Linear programming
Principal component analysis
Scalability
Electric power utilization

ASJC Scopus subject areas

  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this

A Memristor-Based Optimization Framework for Artificial Intelligence Applications. / Liu, Sijia; Wang, Yanzhi; Fardad, Makan; Varshney, Pramod Kumar.

In: IEEE Circuits and Systems Magazine, Vol. 18, No. 1, 8288635, 01.01.2018, p. 29-44.

Research output: Contribution to journalArticle

@article{f7021e4822684755b3a4fb412ba576dd,
title = "A Memristor-Based Optimization Framework for Artificial Intelligence Applications",
abstract = "Memristors have recently received significant attention as device-level components for building a novel generation of computing systems. These devices have many promising features, such as non-volatility, low power consumption, high density, and excellent scalability. The ability to control and modify biasing voltages at memristor terminals make them promising candidates to efficiently perform matrix-vector multiplications and solve systems of linear equations. In this article, we discuss how networks of memristors arranged in crossbar arrays can be used for efficiently solving optimization and machine learning problems. We introduce a new memristor-based optimization framework that combines the computational merits of memristor crossbars with the advantages of an operator splitting method, the alternating direction method of multipliers (ADMM). Here, ADMM helps in splitting a complex optimization problem into subproblems that involve the solution of systems of linear equations. The strength of this framework is shown by applying it to linear programming, quadratic programming, and sparse optimization. In addition to ADMM, implementation of a customized power iteration method for eigenvalue/eigenvector computation using memristor crossbars is discussed. The memristor-based power iteration method can further be applied to principal component analysis. The use of memristor crossbars yields a significant speed-up in computation, and thus, we believe, has the potential to advance optimization and machine learning research in artificial intelligence.",
author = "Sijia Liu and Yanzhi Wang and Makan Fardad and Varshney, {Pramod Kumar}",
year = "2018",
month = "1",
day = "1",
doi = "10.1109/MCAS.2017.2785421",
language = "English (US)",
volume = "18",
pages = "29--44",
journal = "IEEE Circuits and Systems Magazine",
issn = "1531-636X",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

TY - JOUR

T1 - A Memristor-Based Optimization Framework for Artificial Intelligence Applications

AU - Liu, Sijia

AU - Wang, Yanzhi

AU - Fardad, Makan

AU - Varshney, Pramod Kumar

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Memristors have recently received significant attention as device-level components for building a novel generation of computing systems. These devices have many promising features, such as non-volatility, low power consumption, high density, and excellent scalability. The ability to control and modify biasing voltages at memristor terminals make them promising candidates to efficiently perform matrix-vector multiplications and solve systems of linear equations. In this article, we discuss how networks of memristors arranged in crossbar arrays can be used for efficiently solving optimization and machine learning problems. We introduce a new memristor-based optimization framework that combines the computational merits of memristor crossbars with the advantages of an operator splitting method, the alternating direction method of multipliers (ADMM). Here, ADMM helps in splitting a complex optimization problem into subproblems that involve the solution of systems of linear equations. The strength of this framework is shown by applying it to linear programming, quadratic programming, and sparse optimization. In addition to ADMM, implementation of a customized power iteration method for eigenvalue/eigenvector computation using memristor crossbars is discussed. The memristor-based power iteration method can further be applied to principal component analysis. The use of memristor crossbars yields a significant speed-up in computation, and thus, we believe, has the potential to advance optimization and machine learning research in artificial intelligence.

AB - Memristors have recently received significant attention as device-level components for building a novel generation of computing systems. These devices have many promising features, such as non-volatility, low power consumption, high density, and excellent scalability. The ability to control and modify biasing voltages at memristor terminals make them promising candidates to efficiently perform matrix-vector multiplications and solve systems of linear equations. In this article, we discuss how networks of memristors arranged in crossbar arrays can be used for efficiently solving optimization and machine learning problems. We introduce a new memristor-based optimization framework that combines the computational merits of memristor crossbars with the advantages of an operator splitting method, the alternating direction method of multipliers (ADMM). Here, ADMM helps in splitting a complex optimization problem into subproblems that involve the solution of systems of linear equations. The strength of this framework is shown by applying it to linear programming, quadratic programming, and sparse optimization. In addition to ADMM, implementation of a customized power iteration method for eigenvalue/eigenvector computation using memristor crossbars is discussed. The memristor-based power iteration method can further be applied to principal component analysis. The use of memristor crossbars yields a significant speed-up in computation, and thus, we believe, has the potential to advance optimization and machine learning research in artificial intelligence.

UR - http://www.scopus.com/inward/record.url?scp=85042135917&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85042135917&partnerID=8YFLogxK

U2 - 10.1109/MCAS.2017.2785421

DO - 10.1109/MCAS.2017.2785421

M3 - Article

VL - 18

SP - 29

EP - 44

JO - IEEE Circuits and Systems Magazine

JF - IEEE Circuits and Systems Magazine

SN - 1531-636X

IS - 1

M1 - 8288635

ER -