Tackling multimodal device distributions in inverse photonic design using invertible neural networks
We show how conditional generative neural networks can be used to efficiently find nanophotonic devices with desired properties, also known as inverse photonic design. Machine learning has emerged as a promising approach to overcome limitations imposed by the dimensionality and topology of the param...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IOP Publishing
2023-01-01
|
Series: | Machine Learning: Science and Technology |
Subjects: | |
Online Access: | https://doi.org/10.1088/2632-2153/acd619 |
_version_ | 1797815165570252800 |
---|---|
author | Michel Frising Jorge Bravo-Abad Ferry Prins |
author_facet | Michel Frising Jorge Bravo-Abad Ferry Prins |
author_sort | Michel Frising |
collection | DOAJ |
description | We show how conditional generative neural networks can be used to efficiently find nanophotonic devices with desired properties, also known as inverse photonic design. Machine learning has emerged as a promising approach to overcome limitations imposed by the dimensionality and topology of the parameter space. Importantly, traditional optimization routines assume an invertible mapping between the design parameters and response. However, different designs may have comparable or even identical performance confusing the optimization algorithm when performing inverse design. Our generative modeling approach provides the full distribution of possible solutions to the inverse design problem, including multiple solutions. We compare a commonly used conditional variational autoencoder (cVAE) and a conditional invertible neural network (cINN) on a proof-of-principle nanophotonic problem, consisting in tailoring the transmission spectrum trough a metallic film milled by subwavelength indentations. We show how cINNs have superior flexibility compared to cVAEs when dealing with multimodal device distributions. |
first_indexed | 2024-03-13T08:18:35Z |
format | Article |
id | doaj.art-4f07158efaae4467a4fc4f7a7b92f6fa |
institution | Directory Open Access Journal |
issn | 2632-2153 |
language | English |
last_indexed | 2024-03-13T08:18:35Z |
publishDate | 2023-01-01 |
publisher | IOP Publishing |
record_format | Article |
series | Machine Learning: Science and Technology |
spelling | doaj.art-4f07158efaae4467a4fc4f7a7b92f6fa2023-05-31T11:16:52ZengIOP PublishingMachine Learning: Science and Technology2632-21532023-01-014202LT0210.1088/2632-2153/acd619Tackling multimodal device distributions in inverse photonic design using invertible neural networksMichel Frising0https://orcid.org/0000-0003-3725-4824Jorge Bravo-Abad1https://orcid.org/0000-0001-6876-8022Ferry Prins2https://orcid.org/0000-0001-7605-1566Condensed Matter Physics Center (IFIMAC) and Department of Condensed Matter Physics, Autonomous University of Madrid , 28049 Madrid, SpainCondensed Matter Physics Center (IFIMAC) and Department of Theoretical Condensed Matter Physics, Autonomous University of Madrid , 28049 Madrid, SpainCondensed Matter Physics Center (IFIMAC) and Department of Condensed Matter Physics, Autonomous University of Madrid , 28049 Madrid, SpainWe show how conditional generative neural networks can be used to efficiently find nanophotonic devices with desired properties, also known as inverse photonic design. Machine learning has emerged as a promising approach to overcome limitations imposed by the dimensionality and topology of the parameter space. Importantly, traditional optimization routines assume an invertible mapping between the design parameters and response. However, different designs may have comparable or even identical performance confusing the optimization algorithm when performing inverse design. Our generative modeling approach provides the full distribution of possible solutions to the inverse design problem, including multiple solutions. We compare a commonly used conditional variational autoencoder (cVAE) and a conditional invertible neural network (cINN) on a proof-of-principle nanophotonic problem, consisting in tailoring the transmission spectrum trough a metallic film milled by subwavelength indentations. We show how cINNs have superior flexibility compared to cVAEs when dealing with multimodal device distributions.https://doi.org/10.1088/2632-2153/acd619nanotechnologyinverse designmultimodalitymachine learninginvertible neural networksnanophotonics |
spellingShingle | Michel Frising Jorge Bravo-Abad Ferry Prins Tackling multimodal device distributions in inverse photonic design using invertible neural networks Machine Learning: Science and Technology nanotechnology inverse design multimodality machine learning invertible neural networks nanophotonics |
title | Tackling multimodal device distributions in inverse photonic design using invertible neural networks |
title_full | Tackling multimodal device distributions in inverse photonic design using invertible neural networks |
title_fullStr | Tackling multimodal device distributions in inverse photonic design using invertible neural networks |
title_full_unstemmed | Tackling multimodal device distributions in inverse photonic design using invertible neural networks |
title_short | Tackling multimodal device distributions in inverse photonic design using invertible neural networks |
title_sort | tackling multimodal device distributions in inverse photonic design using invertible neural networks |
topic | nanotechnology inverse design multimodality machine learning invertible neural networks nanophotonics |
url | https://doi.org/10.1088/2632-2153/acd619 |
work_keys_str_mv | AT michelfrising tacklingmultimodaldevicedistributionsininversephotonicdesignusinginvertibleneuralnetworks AT jorgebravoabad tacklingmultimodaldevicedistributionsininversephotonicdesignusinginvertibleneuralnetworks AT ferryprins tacklingmultimodaldevicedistributionsininversephotonicdesignusinginvertibleneuralnetworks |