skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Deep learning for accelerated all-dielectric metasurface design

Abstract

Deep learning has risen to the forefront of many fields in recent years, overcoming challenges previously considered intractable with conventional means. Materials discovery and optimization is one such field, but significant challenges remain, including the requirement of large labeled datasets and one-to-many mapping that arises in solving the inverse problem. Here we demonstrate modeling of complex all-dielectric metasurface systems with deep neural networks, using both the metasurface geometry and knowledge of the underlying physics as inputs. Our deep learning network is highly accurate, achieving an average mean square error of only 1.16 × 10-3 and is over five orders of magnitude faster than conventional electromagnetic simulation software. We further develop a novel method to solve the inverse modeling problem, termed fast forward dictionary search (FFDS), which offers tremendous controls to the designer and only requires an accurate forward neural network model. These techniques significantly increase the viability of more complex all-dielectric metasurface designs and provide opportunities for the future of tailored light matter interactions.

Authors:
; ; ; ORCiD logo
Publication Date:
Research Org.:
Duke Univ., Durham, NC (United States)
Sponsoring Org.:
USDOE Office of Science (SC); Alfred P. Sloan Foundation
OSTI Identifier:
1562177
Alternate Identifier(s):
OSTI ID: 1612154
Grant/Contract Number:  
SC0014372
Resource Type:
Published Article
Journal Name:
Optics Express
Additional Journal Information:
Journal Name: Optics Express Journal Volume: 27 Journal Issue: 20; Journal ID: ISSN 1094-4087
Publisher:
Optical Society of America (OSA)
Country of Publication:
United States
Language:
English
Subject:
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSICS; Optics

Citation Formats

Nadell, Christian C., Huang, Bohao, Malof, Jordan M., and Padilla, Willie J. Deep learning for accelerated all-dielectric metasurface design. United States: N. p., 2019. Web. doi:10.1364/OE.27.027523.
Nadell, Christian C., Huang, Bohao, Malof, Jordan M., & Padilla, Willie J. Deep learning for accelerated all-dielectric metasurface design. United States. doi:10.1364/OE.27.027523.
Nadell, Christian C., Huang, Bohao, Malof, Jordan M., and Padilla, Willie J. Mon . "Deep learning for accelerated all-dielectric metasurface design". United States. doi:10.1364/OE.27.027523.
@article{osti_1562177,
title = {Deep learning for accelerated all-dielectric metasurface design},
author = {Nadell, Christian C. and Huang, Bohao and Malof, Jordan M. and Padilla, Willie J.},
abstractNote = {Deep learning has risen to the forefront of many fields in recent years, overcoming challenges previously considered intractable with conventional means. Materials discovery and optimization is one such field, but significant challenges remain, including the requirement of large labeled datasets and one-to-many mapping that arises in solving the inverse problem. Here we demonstrate modeling of complex all-dielectric metasurface systems with deep neural networks, using both the metasurface geometry and knowledge of the underlying physics as inputs. Our deep learning network is highly accurate, achieving an average mean square error of only 1.16 × 10-3 and is over five orders of magnitude faster than conventional electromagnetic simulation software. We further develop a novel method to solve the inverse modeling problem, termed fast forward dictionary search (FFDS), which offers tremendous controls to the designer and only requires an accurate forward neural network model. These techniques significantly increase the viability of more complex all-dielectric metasurface designs and provide opportunities for the future of tailored light matter interactions.},
doi = {10.1364/OE.27.027523},
journal = {Optics Express},
number = 20,
volume = 27,
place = {United States},
year = {2019},
month = {9}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record
DOI: 10.1364/OE.27.027523

Citation Metrics:
Cited by: 19 works
Citation information provided by
Web of Science

Save / Share:

Works referenced in this record:

Active Tuning of All-Dielectric Metasurfaces
journal, February 2015


Scalable Nearest Neighbor Algorithms for High Dimensional Data
journal, November 2014

  • Muja, Marius; Lowe, David G.
  • IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 36, Issue 11
  • DOI: 10.1109/TPAMI.2014.2321376

Experimental realization of a terahertz all-dielectric metasurface absorber
journal, January 2017

  • Liu, Xinyu; Fan, Kebin; Shadrivov, Ilya V.
  • Optics Express, Vol. 25, Issue 1
  • DOI: 10.1364/OE.25.000191

Taming the Blackbody with Infrared Metamaterials as Selective Thermal Emitters
journal, July 2011


Invited Article: Narrowband terahertz bandpass filters employing stacked bilayer metasurface antireflection structures
journal, May 2018

  • Chang, Chun-Chieh; Huang, Li; Nogan, John
  • APL Photonics, Vol. 3, Issue 5
  • DOI: 10.1063/1.5003984

Deep learning
journal, May 2015

  • LeCun, Yann; Bengio, Yoshua; Hinton, Geoffrey
  • Nature, Vol. 521, Issue 7553
  • DOI: 10.1038/nature14539

All-Dielectric Full-Color Printing with TiO 2 Metasurfaces
journal, March 2017


Multilayer feedforward networks are universal approximators
journal, January 1989


Training Deep Neural Networks for the Inverse Design of Nanophotonic Structures
journal, January 2018


Nanophotonic particle simulation and inverse design using artificial neural networks
journal, June 2018


Neural Network Inverse Modeling and Applications to Microwave Filter Design
journal, April 2008


All-dielectric metasurface absorbers for uncooled terahertz imaging
journal, January 2017


Dielectric Resonator Reflectarray as High-Efficiency Nonuniform Terahertz Metasurface
journal, April 2016


Deep-Learning-Enabled On-Demand Design of Chiral Metamaterials
journal, May 2018


Degenerate critical coupling in all-dielectric metasurface absorbers
journal, January 2017

  • Ming, Xianshun; Liu, Xinyu; Sun, Liqun
  • Optics Express, Vol. 25, Issue 20
  • DOI: 10.1364/OE.25.024658

Why Does Deep and Cheap Learning Work So Well?
journal, July 2017

  • Lin, Henry W.; Tegmark, Max; Rolnick, David
  • Journal of Statistical Physics, Vol. 168, Issue 6
  • DOI: 10.1007/s10955-017-1836-5

Migrating Knowledge between Physical Scenarios Based on Artificial Neural Networks
journal, April 2019


Approximation by superpositions of a sigmoidal function
journal, December 1989

  • Cybenko, G.
  • Mathematics of Control, Signals, and Systems, Vol. 2, Issue 4
  • DOI: 10.1007/BF02551274

Plasmonic nanostructure design and characterization via Deep Learning
journal, September 2018

  • Malkiel, Itzik; Mrejen, Michael; Nagler, Achiya
  • Light: Science & Applications, Vol. 7, Issue 1
  • DOI: 10.1038/s41377-018-0060-7