scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

Metamaterial Design Using Distributed Neural Network (DiNN) Approach

TL;DR: A novel Distributed Neural Network (DiNN) approach for the design of Metamaterial which doesn’t require a high computational facility, with a faster training process and high accuracy with 95% of mean square error (MSE) below $4.6x10^{-4}$.
Abstract: We propose a novel Distributed Neural Network (DiNN) approach for the design of Metamaterial which doesn’t require a high computational facility, with a faster training process and high accuracy with 95% of mean square error (MSE) below $4.6x10^{-4}$.
References
More filters
Proceedings ArticleDOI
19 Oct 2008
TL;DR: In this paper, the authors demonstrate THz metamaterials exhibiting either amplitude control via carrier injection or depletion in the active semiconductor substrate or frequency control via photoexcitation of carriers into active semiconducting materials incorporated into the sub-wavelength metammaterial structure.
Abstract: We demonstrate THz metamaterials exhibiting either amplitude control, via carrier injection or depletion in the active semiconductor substrate or frequency control, via photoexcitation of carriers into active semiconducting materials incorporated into the sub-wavelength metamaterial structure.

679 citations

Journal ArticleDOI
TL;DR: A novel method to solve the inverse modeling problem, termed fast forward dictionary search (FFDS), is developed, which offers tremendous controls to the designer and only requires an accurate forward neural network model.
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.

279 citations

Posted Content
TL;DR: In this article, an inverse design and experimentally demonstrated a three-channel wavelength demultiplexer with 40 nm spacing (1500 nm, 1540 nm, and 1580 nm) with a footprint of 24.75 $\mu\mathrm{m}^2
Abstract: In wavelength division multiplexing (WDM) schemes, splitters must be used to combine and separate different wavelengths. Conventional splitters are fairly large with footprints in hundreds to thousands of square microns, and experimentally-demonstrated MMI-based and inverse-designed ultra-compact splitters operate with only two channels and large channel spacing ($>$100 nm). Here we inverse design and experimentally demonstrate a three-channel wavelength demultiplexer with 40 nm spacing (1500 nm, 1540 nm, and 1580 nm) with a footprint of 24.75 $\mu\mathrm{m}^2$. The splitter has a simulated peak insertion loss of -1.55 dB with under -15 dB crosstalk and a measured peak insertion loss of -2.29 dB with under -10.7 dB crosstalk.

52 citations

Posted Content
Yang Deng1, Simiao Ren1, Kebin Fan1, Jordan M. Malof1, Willie J. Padilla1 
TL;DR: This work proposes and demonstrates a method capable of finding accurate solutions to ill-posed inverse problems, where the conditions of existence and uniqueness are violated, and shows how the neural-adjoint method can intelligently grow the design search space to include designs that increasingly and accurately approximate the desired scattering response.
Abstract: All-dielectric metasurfaces exhibit exotic electromagnetic responses, similar to those obtained with metal-based metamaterials. Research in all-dielectric metasurfaces currently uses relatively simple unit-cell designs, but increased geometrical complexity may yield even greater scattering states. Although machine learning has recently been applied to the design of metasurfaces with impressive results, the much more challenging task of finding a geometry that yields the desired spectra remains largely unsolved. We explore and adapt a recent deep learning approach -- termed neural-adjoint -- and find it is capable of accurately and efficiently estimating complex geometry needed to yield a targeted frequency-dependent scattering. We also show how the neural-adjoint method can intelligently grow the design search space to include designs that increasingly and accurately approximate the desired scattering response. The neural-adjoint method is not restricted to the case demonstrated and may be applied to plasmonics, photonic bandgap, and other structured material systems.

22 citations

Journal ArticleDOI
TL;DR: A deep neural network is built for on-demand design of metamaterials and indicates that using deep learning to train the data, the trained model can more accurately guide the design of the structure, thereby speeding up the design process.
Abstract: The introduction of “metamaterials” has had a profound impact on several fields, including electromagnetics. Designing a metamaterial’s structure on demand, however, is still an extremely time-consuming process. As an efficient machine learning method, deep learning has been widely used for data classification and regression in recent years and in fact shown good generalization performance. We have built a deep neural network for on-demand design. With the required reflectance as input, the parameters of the structure are automatically calculated and then output to achieve the purpose of designing on demand. Our network has achieved low mean square errors (MSE), with MSE of 0.005 on both the training and test sets. The results indicate that using deep learning to train the data, the trained model can more accurately guide the design of the structure, thereby speeding up the design process. Compared with the traditional design process, using deep learning to guide the design of metamaterials can achieve faster, more accurate, and more convenient purposes.

18 citations