Inverse design of grating couplers using the policy gradient method from reinforcement learning

We present a proof-of-concept technique for the inverse design of electromagnetic devices motivated by the policy gradient method in reinforcement learning, named PHORCED (PHotonic Optimization using REINFORCE Criteria for Enhanced Design). This technique uses a probabilistic generative neural netwo...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Hooten Sean, Beausoleil Raymond G., Van Vaerenbergh Thomas
Formato:	article
Lenguaje:	EN
Publicado:	De Gruyter 2021
Materias:	adjoint method; deep learning; integrated photonics; inverse design; optimization; reinforcement learning Physics QC1-999
Acceso en línea:	https://doaj.org/article/73c23020fca646c58959e509beee6f4c
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:73c23020fca646c58959e509beee6f4c
record_format	dspace
spelling	oai:doaj.org-article:73c23020fca646c58959e509beee6f4c2021-12-05T14:10:56ZInverse design of grating couplers using the policy gradient method from reinforcement learning2192-861410.1515/nanoph-2021-0332https://doaj.org/article/73c23020fca646c58959e509beee6f4c2021-10-01T00:00:00Zhttps://doi.org/10.1515/nanoph-2021-0332https://doaj.org/toc/2192-8614We present a proof-of-concept technique for the inverse design of electromagnetic devices motivated by the policy gradient method in reinforcement learning, named PHORCED (PHotonic Optimization using REINFORCE Criteria for Enhanced Design). This technique uses a probabilistic generative neural network interfaced with an electromagnetic solver to assist in the design of photonic devices, such as grating couplers. We show that PHORCED obtains better performing grating coupler designs than local gradient-based inverse design via the adjoint method, while potentially providing faster convergence over competing state-of-the-art generative methods. As a further example of the benefits of this method, we implement transfer learning with PHORCED, demonstrating that a neural network trained to optimize 8° grating couplers can then be re-trained on grating couplers with alternate scattering angles while requiring >10× fewer simulations than control cases.Hooten SeanBeausoleil Raymond G.Van Vaerenbergh ThomasDe Gruyterarticleadjoint method; deep learning; integrated photonics; inverse design; optimization; reinforcement learningPhysicsQC1-999ENNanophotonics, Vol 10, Iss 15, Pp 3843-3856 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	adjoint method; deep learning; integrated photonics; inverse design; optimization; reinforcement learning Physics QC1-999
spellingShingle	adjoint method; deep learning; integrated photonics; inverse design; optimization; reinforcement learning Physics QC1-999 Hooten Sean Beausoleil Raymond G. Van Vaerenbergh Thomas Inverse design of grating couplers using the policy gradient method from reinforcement learning
description	We present a proof-of-concept technique for the inverse design of electromagnetic devices motivated by the policy gradient method in reinforcement learning, named PHORCED (PHotonic Optimization using REINFORCE Criteria for Enhanced Design). This technique uses a probabilistic generative neural network interfaced with an electromagnetic solver to assist in the design of photonic devices, such as grating couplers. We show that PHORCED obtains better performing grating coupler designs than local gradient-based inverse design via the adjoint method, while potentially providing faster convergence over competing state-of-the-art generative methods. As a further example of the benefits of this method, we implement transfer learning with PHORCED, demonstrating that a neural network trained to optimize 8° grating couplers can then be re-trained on grating couplers with alternate scattering angles while requiring >10× fewer simulations than control cases.
format	article
author	Hooten Sean Beausoleil Raymond G. Van Vaerenbergh Thomas
author_facet	Hooten Sean Beausoleil Raymond G. Van Vaerenbergh Thomas
author_sort	Hooten Sean
title	Inverse design of grating couplers using the policy gradient method from reinforcement learning
title_short	Inverse design of grating couplers using the policy gradient method from reinforcement learning
title_full	Inverse design of grating couplers using the policy gradient method from reinforcement learning
title_fullStr	Inverse design of grating couplers using the policy gradient method from reinforcement learning
title_full_unstemmed	Inverse design of grating couplers using the policy gradient method from reinforcement learning
title_sort	inverse design of grating couplers using the policy gradient method from reinforcement learning
publisher	De Gruyter
publishDate	2021
url	https://doaj.org/article/73c23020fca646c58959e509beee6f4c
work_keys_str_mv	AT hootensean inversedesignofgratingcouplersusingthepolicygradientmethodfromreinforcementlearning AT beausoleilraymondg inversedesignofgratingcouplersusingthepolicygradientmethodfromreinforcementlearning AT vanvaerenberghthomas inversedesignofgratingcouplersusingthepolicygradientmethodfromreinforcementlearning
_version_	1718371546315096064

Inverse design of grating couplers using the policy gradient method from reinforcement learning

Ejemplares similares