Research Article Vol. 30, No.

2 / 17 Jan 2022 / Optics Express 3076

Machine-learning-assisted inverse design of

scattering enhanced metasurface
H AI L IN , 1,5 J UNJIE H OU , 1 J ING JIN , 1 Y UMEI WANG , 2 R ONGXIN
TANG , 3,6 X INTONG S HI , 1 Y UZE T IAN , 1 AND W EILIN X U 4
1 College of physical science and technology, central china normal university, Wuhan 430000, China
2 China Ship Development and Design Centre, Wuhan 430000, China
3 Institute of Space Science and Technology, Nanchang University, 330021, China
4 School of the Information and Communication, Guilin University of Electronic Technology, Guilin 541004,
China
5 linhai@[Link]
6 rongxint@[Link]

Abstract: The scattering enhancement technique has shown prominent potential in various
regimes such as satellite communication, Radar Cross Section (RCS) camouflage, and remote
sensing. Currently, the scattering enhancement devices based on the metasurface have shown
advantages in light weight and better performance. These metasurfaces always possess complex
structure, it is hard to achieve through the tradition trial-and-error method which relies on the
full-wave numerical simulation. In this paper, a new method combining the machine learning and
the evolution optimization algorithm is proposed to design the metasurface retroreflector (MRF)
for arbitrary direction incident wave. In this method, a predicting model and a generative inverse
design model are constructed and trained, the predicting model is used to evaluate the fitness
of each offspring in the genetic algorithm (GA), the generative model is used to initialize the
first offspring of the GA by inverse generate the MRF based on the requirements of the designer.
With the assistance of these two machine learning models, the evolution optimization algorithm
is employed to find the optimal design of the MRF. This approach enables automatic solution of
electromagnetic inverse design problems and opens the way to facilitate the optimization of other
metadevices.

© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement

1. Introduction
Controlling the scattering is a key subject in theory and application of electromagnetic (EM). If
one can arbitrarily manipulate the scattering field or radar cross section (RCS) of the scatterer, it
will be possible to disguise the object as nothing or others. In the past, the emphasis of research is
focused on RCS reduction for radar stealth or electromagnetic compatibility (EMC) applications.
Since the scattering field depends on the geometry, material properties, and impedance loading
of the object, the traditional manners to reduce the scattering field include shaping of the target
geometry, impedance loading, radar absorbing materials, and active methods [1]. Recently,
another scattering manipulating scheme including invisibility cloak [2], radar illusion [3], carpet
cloak [4,5], and mantle cloak [6] has also been provided based on transformation optics theory.
On the other hand, scattering augmentation has also been widely used in both military and civil
research fields. To name a few, a RCS-augmented scatterer serving as a radar decoy can improve
the survivability of true target [7]. With enhanced retroreflection, the echoes of small vehicles,
such as lifeboats and dinghies might be increased, which can facilitate rescue or navigation safety.
Another urgent demand is how to enhance the backscattering echo of chipless radio-frequency
identification (RFID) tag which will greatly improve the rate of identification for IoT applications.
Unfortunately, relatively less attention is concentrated on the scheme for scattering enhancement.

#448051 [Link]
Journal © 2022 Received 10 Nov 2021; revised 22 Dec 2021; accepted 3 Jan 2022; published 14 Jan 2022
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3077

Retroreflector is a device or surface that reflects incident EM waves back to its incoming
direction by enhancing the backward scattering. So far, devices like corner reflector and Luneburg
lens are implemented to realize retroreflection. However, due to their bulky and nonplanar
structures, these devices inevitably have limitations in miniaturization and integration with other
components. To deal with the above problems, metasurface has been introduced to this field.
In the past decade, metasurface has emerged as one of the most potential candidates for EM wave
control devices, it can provide fascinating capabilities for EM control such as perfect absorption,
anomalous reflection, wavefront shaping, and dispersion engineering [8–12]. Metasurfaces are
typically made of planar dense arrays with subwavelength unit cells. In principle, EM waves can
be manipulated flexibly by controlling the magnitude and phase of the unit cells. In 2014, Cui et
al. proposed a new concept of coding metasurface by arranging the corresponding coding units
as a predesigned coding matrix [13]. The coding units are obtained by encoding the discrete
phases of unit cells in binary numbers. The coding metasurface is of more freedom than the
traditional metasurface, it can form various scattering patterns due to constructive/destructive
interference or phase cancellation for RCS enhancement or reduction in designated directions.
For instance, in [14], Su et al. have proposed a checkerboard metasurface for super-wideband
RCS reduction based on diffuse reflection theory and phase cancellation principle. In [15], Cheng
et al. have proposed a nonperiodic metasurface used for retroreflection of oblique incident wave.
In [16], feng et al. have proposed a method based on phase gradient metasurface to realize RCS
enhancement by reflecting the incident wave to the desired direction.
Currently, the traditional design method of metasurface still relies on the trial-and-error method
and the intuition of the designer [17–20]. With the increasing complexity of metasurface structure,
especially the coding metasurface which has an extremely high complexity, this conventional
approach gradually shows its limitations. The repeated calculations require a lot of computation
resources and effort of the designer. Thus, the optimization algorithm has been applied to the
inverse design of metasurface by some researchers.
The optimization algorithms such as topology optimization [21], genetic algorithm (GA) [22],
and gradient-based algorithm [23] contain iterative searching steps and rely on estimate methods
such as full-wave numerical simulation in each step to produce intermediate results that help
optimize the searching strategy. For instance, in [24], Mahdad Mansouree et al. have proposed
an adjoint-optimization-based design technique to design large-scale metasurface with lower
computational cost. In [25], Lin et al. have demonstrated the topology optimization of freeform
large-area optical metasurfaces in two and three dimensions. The optimization algorithm can
effectively replace the manpower to do the repetitive work, however, the random-search nature
of such stochastic algorithms still causes some problems when the number of optimization
parameters increase. The first problem is that the initialization of the algorithm affects the
optimized result to a great extent. A good initialization may lead to a fast convergence, while a
bad initialization may heavily delay the convergence or even make the optimization not converge.
Another problem is that the optimization algorithm may converge to the local optimal solution
rather than the global optimal solution. The final problem is that the optimized design can only
work in certain circumstances, once the design goal is modified, the design will need to be
re-optimized.
In the past decade, machine learning has been successfully applied in many areas such as
computer vision, natural language processing, speech recognition, and face recognition [26–28].
As one of the most powerful tools of machine learning, deep learning has excellent generalization
ability and generative ability. Taking advantage of its capability for solving non-intuitive problems,
some researchers have proposed different approaches to design the metasurface [29–32]. In [33],
Xu et al. employed a tandem network to obtain a bifunctional metasurface dealing with photons
and phonons. In [34], Iman Sajedian et al. used a double deep Q-learning network (DDQN)
to find the right material type and the optimal geometrical design for metasurface holograms
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3078

to reach high efficiency. In [35], Sensong An et al. presented a conditional generative network
that can generate metasurface designs based on different performance requirements. In the
deep-learning-based method, once the inverse design model is constructed, it can automatically
and immediately generate the target design according to the requirements, which will significantly
facilitate the design process without the manpower to intervene. However, the deep neural
network (DNN) heavily relies on its training data, the generalization ability of the DNN is mostly
determined by the training data. For a simple structure, it is easy to obtain a mass of training
data that the DNN can be easily trained to fulfill its task. Nevertheless, as the complexity of the
problem increases, it will be much more difficult to acquire amounts of valid data to train the
DNN.
In this paper, a deep-learning-assisted genetic algorithm (DLAGA) is proposed to find the
optimal design of the metasurface retroreflector (MRF). As is shown in Fig. 1, this method is
composed of three modules, a generative model named as conditional variational auto-encoder
(CVAE), a predicting neural network (PNN) and an optimization module based on GA. The CVAE
is trained to generate the satisfactory coding matrix of MRF according to the input condition, and
it is employed to guarantee the good initialization of GA. The PNN can predict the monostatic
RCS fastly and accurately, therefore, it is used to evaluate the fitness of each offspring in the
GA. With the assistance of CVAE and PNN, the GA is more likely to find the optimal solution.
We also proposed an efficient data collection method that can obtain valid data rapidly. It can

Fig. 1. Schematic of the machine-learning-assisted method.

 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3079

greatly accelerate the establishment of the training data set for neural networks. To showcase the
performance of the proposed method, we optimized four MRF with excellent performance.

2. Results and discussion

As presented in [36,37], the total reflected energy from the metasurface upon the incidence of a
plane wave is the sum of the reflections from all unit cells. The scattered field pattern can be
represented by the superposition of the initial excitation phase and phase response of each unit
cell [15]:
N
∑︂ N
∑︂
f (θ r ) = [expjkmdsin(θi ) expjkmdsin(θr ) expjϕ(m,n) ] (1)
m=1 n=1

where φ(m, n) represents the phase response of each unit cell (m, n); m and n denote the position of
the unit cell in x-dimension and y-dimension. k is the wave number in free space, d is the distance
between the adjacent element, θ i and θ r are incidence and scattered wave angle, and jkmdsin(θ i )
is the excitation phase related to the location of unit cell. This is a general scattered pattern
equation of the coding metasurface for all possible phase responses with given polarization. To
achieve the desired retroreflection pattern, φ(m, n) should be determined at each position on the
metasurface. Here we proposed a machine-learning-assisted method to find a properly arranged
coding matrix of the metasurface to achieve maximum reflection in the retroreflection direction.

2.1. Data collection method

The direction of the incident wave can be described by angles θ and φ in spherical coordinates.
The RCS of the metasurface are both determined by the incident angle and the phase pattern of
the unit cells. The MRF is arranged as a 16 × 16 array wihch composed of four types of unit
cells. To simply demonstrate the robustness and effectiveness of our inverse design algorithm,
we restrict the incident wave within the plane φ = 0 with θ in the range of 0◦ ∼90◦ , hence, the
arrangement of the unit cells is only change along one direction and remain unchanged in the
other direction. We also assume that the unit cells of the metasurface have a simple structure
which is depicted in Fig. 2. The unit cell is designed as square metallic patch with width w which
is printed on an F4B dielectric substrate. This dielectric substrate has a thickness of h=2mm,
dielectric constant εr =2.65 and loss tangent tanδ=0.001. The periodicity of the unit cell is
p=5mm. A metal ground is placed on the bottom of the dielectric substrate to block the incident
wave from transmission. From the initial simulation study, we found that when the designed
patch widths are 4.9, 4.4, 4 and 0.5 mm, the reflective amplitude of four elements is close to 1 at
11 GHz while their phase difference is approximately 90◦ as shown in Fig. 2(c) and Fig. 2(d).
According to [13], these four types of unit cells can be used to define a 2-bit coding metasurface
with coding sequence ’1’, ’2’, ’3’, ’4’ respectively. Although the unit cells are only change along
one direction which means the MRF can be described as a 16 × 1 coding sequence, there are
still 416 possible coding matrices which form an extremely large solution space. To obtain a
well-trained deep learning model, effective data collection is quite demanded.
Data collection has always been a sore point of the machine-learning-based method in the
design of metasurfaces. In [38], Tao shan et al. used GA to compute the element codes of
metasurface based on the form of the scattering wave. However, GA can only get one batch of
labeled data through one optimization process, which is inefficient owing to the computation
time wasted on the repeated samples.
To address the above-mentioned problems, an efficient data collection method is proposed based
on the customized GA which is named as memory genetic algorithm (MGA). MGA is customized
to select the qualified samples which have significant enhanced effect on retroreflection. The
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3080

Fig. 2. The 2-bit coding metasurface. (a) An example of the coding metasurface. The ’1’,
’2’, ’3’ and ’4’ elements (from left to right) are realized by square metallic patches with
different sizes.

enhanced effect of retroreflection is assessed through the monostatic RCS of the metasurface, the
figure of merit (FoM) of the metasurface is described as:

FoM = W1 ∗ RCSr + W2 ∗ RCSm (2)

where RCSm represents the maximum value of the monostatic RCS, RCSr represents the sum
of RCS values which neighborhood of the RCSm and has a difference with RCSm less than 3dB.
The W1 and W2 are the weighting coefficients which are used to balance the emphasis of the
optimization, the value of both W1 and W2 is range from 0 to 1, and the sum of W1 and W2
is always 1. During the optimization process, by estimate the FoM of each sample, the MGA
will decide whether the sample is qualified. In the continual optimization process, the MGA is
employed to obtain enough MRF samples.
The flowchart of the MGA is shown in Fig. 3(b). As is shown in the figure, we assigned two
datasets named total dataset (TDS) and the qualified dataset (QDS). TDS is used to store all
the samples produced during the optimization process, and QDS is used to store the qualified
samples. The procedure of MGA is as follows. A population of N chromosomes are initialized
by randomly selecting N samples from the QDS. The fitness of each chromosome is computed by
the simulation software, and the qualified chromosomes are stored in the QDS. If the number of
generations is less than 20, two chromosomes will be selected according to the fitness value. The
single-point crossover operator is applied to the selected chromosomes to produce an offspring.
Thereafter, uniform mutation operator is applied to the produced offspring. The mutated offspring
are placed in the new population. The selection, crossover, and mutation operations will be
repeated on the current population until the new population is complete. After each offspring is
produced, the MGA will check each sample in this offspring, if the sample is in the TDS, the
corresponding fitness will be fetched from TDS directly and pass the simulation step. The MGA
will continue producing data until the amount of data is sufficient. Compared to the classical GA,
the MGA has the advantage of avoiding repetitive evaluations of the same samples, which can
effectively accelerate the speed of optimization process and data generation.
Through MGA, 60000 samples of the coding metasurface have been collected, which all
of them will be used to train the PNN. Among all the collected data, only 16000 of them are
qualified and assigned as the training data set of the CVAE.
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3081

Fig. 3. Schematic of the GA and MGA.

2.2. Forward prediction

Since the above defined FoM is calculated from the RCS of metasurface, iteratively RCS
calculation will slow down the whole optimization process. To accelerate the evaluation process
during the optimization, a surrogate model named as PNN is constructed. The PNN is a fast
data-driven DNN that can predict the RCS of the metasurface according to its coding matrix, the
high accuracy and short response time make it a perfect surrogate model for numerical simulation
in the inverse design. DNN is an artificial neural network with many hidden layers, where each
hidden layer consists of multiple hidden neurons. It has been recognized to be powerful at
representing very complex input-output relationships [39,40]. Since the relation between the
metasurface and its RCS is complicated, it is a good choice to use DNN to uncover the hidden
relation between the metasurface and its RCS.
The predicting process is shown in Fig. 4(a). As shown in the figure, based on the design, the
meatsurface is encoded as the coding matrix which is 16-dimensional vector, it is fed into the PNN
as the inputs. After the calculation of the PNN, it will output the corresponding predicted RCS
under 121 sampling angles(−60◦ −60◦ ). The detail of the proposed PNN is shown in Fig. 5(d).
Theoretically, the fitting performance of the neural network is proportional to the number of
hidden layers and neurons [39,41], but as the network deepens, a series of problems will arise,
such as gradient exploding or gradient vanishing, which may cause invalid calculations and
other wastes of computing resources, and greatly increase the time for model training. Different
numbers of hidden layers and neurons are tried in the experiment to achieve the best performance
of the model without increasing the computational burden. In order to accelerate the training
process of PNN, the batch normalization is added behind the input layer and each hidden layer
of the neural network. The batch normalization can effectively avoid the vanishing gradient
problem and increase the generalization ability of the neural network [42]. LeakyReLU activation
function is used as the activation function for neurons in the hidden layer. For the PNN training
process, the least-squares error function, which is also known as mean squared error (MSE), is
used as the loss function. The MSE is expressed as
N
1 ∑︂
LossPNN = (RCSpre − RCSsim )2 (3)
N i=1
Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3082

Fig. 4. (a) Schematic of the predicting process. (b) The MSELoss of the training process.

Fig. 5. (a) Structure of AE. (b) Structure of VAE. (c) Structure of the proposed CVAE and
its parametric form of input and output.
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3083

During the training process, among all the collected data, 55000 samples are assigned as the
training data set, while the remaining 5000 samples are assigned as validation data set. The
validation data set does not directly participate in the training process of the PNN, but it is used
to monitor the quality of the PNN model during training and to determine the stop criteria for the
training process. It can also help adjust the hyper-parameters of the training. The learning rate
is 0.00001, the batch size is 128, and the epoch number is 50. After the training, to verify the
validity and accuracy of PNN, 300 sets of coding matrices are sent to the PNN and full-wave
simulation software at the same time. Note that those samples are generated randomly and
selected to make sure that they are not included in the training data set for training the PNN, the
MSE between the simulation results and the predicted results is 3.24. And the corresponding
loss-epoch number graphs is shown in Fig. 4(b). The corresponding average error rate of the
predicting result is only 3.9%, which is computed by
N
1 ∑︂ RCSpre − RCSsim
ErrorPNN = (| |) (4)
N i=1 RCSsim

To intuitively demonstrate the accuracy of the PNN, several comparative results are shown in
Fig. 4(c). These comparisons show that excellent matches have been achieved between the
simulation results and the PNN-predicted results. Given the vital role the PNN plays in our
machine-learning-assisted approach, such agreement is critical to the inverse design of MRF.

2.3. Conditional variational auto-encoder

To generate the target MRF according to the requirement, a CVAE is constructed and trained as
the generative model. The CVAE is a variant of the variational auto-encoder (VAE) which is
originated from auto-encoder (AE). The general idea of AE is to train an encoder and a decoder
to learn the best encoding-decoding scheme using an iterative optimization process. The general
structure of AE is depicted in Fig. 5(a). A VAE can be treated as an AE whose training is
regularized to avoid overfitting and ensure good properties of the latent space, which enable
the generative ability. Instead of encoding the input as a single point in the latent space, VAE
encodes the input as a distribution over the latent space. Compared to the AE, the VAE has the
ability to generate contents. However, the content generated by the VAE is random, which means
that we cannot control the contents generated by the VAE. To generate controllable content, we
propose a CVAE based on the VAE. In the proposed CVAE, conditions are added in the input of
the encoder and decoder. With the assistance of the added conditions, the CVAE can produce
specified samples based on different requirements.
The details of the DNNs for the CVAE’s encoder and decoder are shown in Fig. 5(d), the
structure of the CVAE is determined after multiple experiments to make sure its good performance
and low computational burden. To design the MRF working in arbitrary direction, the half-power
angle range is treated as the enhancement direction of the MRF. The half-power angle range
represents the angle range between the upper angle where the RCS value is 3 dB down from
maximum and the lower angle where RCS value is 3 dB down. The parametric form of the
input and output of the CVAE is shown in Fig. 5(c). The half-power angle range is assigned as
the condition term that is combined with the coding matrix fed into the encoder. The output of
the encoder is the mean and covariance tensor that describes the standard normal distribution.
Based on the mean and covariance tensor, a latent encoded vector can be obtained through the
reparameterize method [43], then the latent encoded vector and the condition are input to the
decoder together. The decoder tries to restore the management sequence of unit cells in the
metasurface. The loss function used to train the CVAE is defined as

LossCVAE = Lossrec + LossKLD (5)

 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3084

The loss function in Eq.(5) is composed of a reconstruction term and a regularization term.
The reconstruction term is expressed as the MSE between the input coding matrix and the
reconstructed coding matrix output by the decoder. Specifically, it is defined as
Lossrec = |x − ˆ︁
x| 2 (6)
This term tends to make the encoding-decoding scheme as performant as possible. The
regularization term is expressed as the Kulback-Leibler (KL) divergence between the returned
distribution and the standard Gaussian distribution. It is defined as
LossKLD = KL[N(µ, ϵ), N(0, 1)] (7)
This term tries to regularize the organization of the latent space by making the distributions
returned by the encoder close to a standard normal distribution. In the training process of a
CVAE, the loss function formulated in Eq.(5) is minimized. After the training, the generative
ability of CVAE is evaluated by employing the CVAE to generate the MRF that works in the
required direction. The generated MRF is then verified by both numerical simulation and the
PNN. Results of the simulation and the PNN exhibit obvious enhancement of the retroreflection
in the target direction, and the results for several cases have been shown in Fig. 6. The whole
design process of these metasurfaces only takes us less than a minute. Even consider the time
consuming on the data collection and training, this deep-learning-based inverse design method is
still an efficient approach for the design of the metasurface. Not to mention this model can also be
used to build an intelligent MRF which can realize real-time control and arbitrary direction of the
retroreflection enhancement. As described in [13], the programmable metasurface is composed
of programmable unit cell where the phase of each unit cell can be controlled by the digital signal
in real-time. By combining the well-trained CVAE with the programmable MRF, the intelligent
MRF can be realized to generate the required design of the MRF in real-time.

Fig. 6. The predicting and simulation results of the generated design, the blue curve
represents the simulation result, the red curve represents the predicting result of the PNN,
the target angle range and the actual angle range of the design are shown in the upper left of
each inset, the actual angle range is covered by the yellow area.

2.4. Deep learning assisted global optimization

The MRFs generated by CVAE have shown obvious enhancement of the retroreflection, but the
performance of the MRFs generated by CVAE is limited by the training data. In the application
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3085

of retroreflector, a better performance is always necessary, if the design of MRF with the
optimal performance is required, the CVAE can not guarantee the MRF has the best performance.
Therefore, a method that can break through the limitation of training data is highly desired.
To achieve the global optimal design of the metasurface, we proposed the DLAGA method.
The schematic of this method is demonstrated in Fig. 7(a). In this method, the initial population
of the GA is produced by the well-trained CVAE according to the requirement. This brings
two advantages, it is assured that the enhancement direction of all initial population produced
by CVAE can match the requirement. In this way, the offspring of the initial generation will
easily inherit this feature. Secondly, the initial population produced by the CVAE will have better
fitness compared to completely random initialization. As the initial population has high fitness,
their offspring are more likely to have higher fitness. To understand this from the aspect of the
solution space, it can be considered that the CVAE has compressed the searching space to enable
the GA to achieve the global optimization. To accelerate the evaluation of each offspring, at the
evaluation step, the PNN is employed to quickly predict the RCS of the offspring and pass it to
the evaluation agent of GA. After the evaluation of each generation, the algorithm will decide
whether the optimal solution has been found. The sign of the optimal solution has been found is
that the sample with the highest fitness keeps invariant for a certain generation. If the GA finds
a satisfactory solution, the optimal solution will be cross-validated by the full-wave numerical

Fig. 7. The design schematic and the optimal results. (a) The schematic of the DLAGA
method. (b) The distribution of the solution space for GA and DLAGA. (c) The optimal
results and their corresponding coding matrix. (d)The retroreflection enhanced effect of the
above optimal results, the blue curve represents the RCS of the metal with equivalent size, the
black curve represents the simulation results, the red curve represents the predicted results
of the PNN, the target direction and the actual angle range of the design which covered by
the yellow area is shown in the upper left of each inset.
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3086

simulation, otherwise, a new offspring will be produced and sent to the PNN. This operation is
repeated until the optimal solution is found or this optimization reaches the max iteration number.
To show the validity of DLAGA, it has been employed to design several MRFs that work
in different directions. The population size is set to 20, and the optimal solution will be the
sample with the highest fitness keeps invariant for 20 generations. In Fig. 7(c,d), four optimized
metasurfaces working in different directions and the enhanced effect of their retroreflection have
been demonstrated. Further more, We have optimized 480 MRFs through both DLAGA and
simple GA under the same condition, and the purpose is to maximize the FoM of MRFs. So
we compared the RCSr and the RCSm of these MRFs. The distribution of the optimized results
for simple GA and DLAGA is demonstrate in the RCSm -RCSr scatter diagram which is shown
in Fig. 7(b). The x-axis of the scatter diagram is RCSr and the y-axis is RCSm , the red points
represent the MRFs optimized by the DLAGA and the black points represent the MRFs optimized
by the GA. The results optimized by DLAGA mainly distributed in the upper left corner of the
scatter diagram and the results optimized by simple GA is distributed dispersively in the scatter
diagram. It means the results optimized by DLAGA generally have better performance compared
to the results optimized by simple GA. And among all the optimized MRFs, the best samples are
optimized by the DLAGA which means DLAGA is more likely to find the optimal solution of the
MRFs.

3. Conclusion
In this paper, we have proposed an arbitrary MRF design method. A machine-learning-assisted
global optimization approach has been proposed to search the optimal design of the metasurface
by utilizing the evolution optimization algorithm on the searching space compressed by the
CVAE. A prediction model and a generative model are constructed and trained to uncover the
relationship between the metasurface and their RCS. We also modified the GA and proposed an
effective data collection method, which can significantly accelerate the construction of the data
set.
Compared with the traditional method mentioned previously, our inverse design method avoids
the vast full-wave numerical simulation and realizes the automatically on-demand design of MRF
in few seconds. We have also shown that by applying the optimization algorithm directly on
the basis of the compressed design space, it is possible to achieve efficient optimization of the
metasurfaces with high performance.
The work we have done in this paper has proved the potential of this machine-learning-assisted
optimization method in future work, and there remain some areas for future research to augment
the utility of this method. Firstly, add more constraints to the samples generated by the MGA,
this may enhance the generative ability of the CVAE. Secondly, applying the design scheme
to higher dimensions that will increase the degrees of freedom to optimize the objectives at
different azimuths which enable the coding metasurface to control the enhanced direction in a
larger region.
We also expect that this machine-learning-assisted optimization design method can be applied
to not only the coding metasurface but other EM devices, such as antennas, integrated optical
circuit devices. This method can also extend to optical frequencies for plasmonic or dielectric
materials. This can be a possible future direction of research.
Funding. open fund of China Ship Development and Design Centre (XM0120190196); National Natural Science
Foundation of China (41974195); Guangxi Key Laboratory of Wireless Wideband Communication and Signal Processing,
Guilin University of Electronic Technology (GXKL06190202); Fundamental Research Funds for the Central Universities
(CCNU19TS073, CCNU20GF004).
Disclosures. The authors declare no conflicts of interest.
Data availability. Data underlying the results presented in this paper are not publicly available at this time but may
be obtained from the authors upon reasonable request.
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3087

References
1. E. F. Knott, J. F. Schaeffer, and M. T. Tulley, Radar cross section (SciTech Publishing, 2004).
2. J. B. Pendry, D. Schurig, and D. R. Smith, “Controlling electromagnetic fields,” Science 312(5781), 1780–1782
(2006).
3. W. X. Jiang, C.-W. Qiu, T. Han, S. Zhang, and T. J. Cui, “Creation of ghost illusions using wave dynamics in
metamaterials,” Adv. Funct. Mater. 23(32), 4028–4034 (2013).
4. R. Liu, C. Ji, J. Mock, J. Chin, T. Cui, and D. Smith, “Broadband ground-plane cloak,” Science 323(5912), 366–369
(2009).
5. H. F. Ma and T. J. Cui, “Three-dimensional broadband ground-plane cloak made of metamaterials,” Nat. Commun.
1(1), 21 (2010).
6. A. Alù, “Mantle cloak: Invisibility induced by a surface,” Phys. Rev. B 80(24), 245115 (2009).
7. R. Y. Mianroodi, H. Heidar, and H. M. Armaki, “Expandable shipboard decoy including adequate rcs by using
trihedral corner reflectors,” IET Sci. Meas. & Technol. 10(5), 485–491 (2016).
8. N. I. Landy, S. Sajuyigbe, J. J. Mock, D. R. Smith, and W. J. Padilla, “Perfect metamaterial absorber,” Phys. Rev. Lett.
100(20), 207402 (2008).
9. M. Rahmanzadeh and A. Khavasi, “Perfect anomalous reflection using a compound metallic metagrating,” Opt.
Express 28(11), 16439–16452 (2020).
10. M. Jang, Y. Horie, A. Shibukawa, J. Brake, Y. Liu, S. M. Kamali, A. Arbabi, H. Ruan, A. Faraon, and C. Yang,
“Wavefront shaping with disorder-engineered metasurfaces,” Nat. Photonics 12(2), 84–90 (2018).
11. Y. Hu, X. Wang, X. Luo, X. Ou, L. Li, Y. Chen, P. Yang, S. Wang, and H. Duan, “All-dielectric metasurfaces for
polarization manipulation: principles and emerging applications,” Nanophotonics 9(12), 3755–3780 (2020).
12. W. T. Chen, A. Y. Zhu, J. Sisler, V. Sanjeev, E. Lee, and F. Capasso, “Dispersion engineering of metasurfaces and its
applications in the visible,” in 2018 Conference on Lasers and Electro-Optics (CLEO), (2018), pp. 1–2.
13. T. J. Cui, M. Q. Qi, X. Wan, J. Zhao, and Q. Cheng, “Coding metamaterials, digital metamaterials and programmable
metamaterials,” Light: Sci. Appl. 3(10), e218 (2014).
14. S. Jianxun, L. Yao, L. Jiayi, Y. Yaoqing, L. Zengrui, and S. Jiming, “A novel checkerboard metasurface based on
optimized multielement phase cancellation for superwideband rcs reduction,” IEEE Trans. Antennas Propag. 66(12),
7091–7099 (2018).
15. C. Tao and T. Itoh, “Nonperiodic metasurfaces for retroreflection of te/tm and circularly polarized waves,” IEEE
Trans. Antennas Propag. 68(8), 6193–6203 (2020).
16. F. MaoChang, L. YongFeng, Z. JieQiu, W. JiaFu, W. Chao, M. Hua, and Q. ShaoBo, “Research of a wide-angle
backscattering enhancement metasurface,” Acta Phys. Sin. 67(19), 198101 (2018).
17. A. Arbabi, Y. Horie, M. Bagheri, and A. Faraon, “Dielectric metasurfaces for complete control of phase and
polarization with subwavelength spatial resolution and high transmission,” Nat. Nanotechnol. 10(11), 937–943
(2015).
18. N. Yu and F. Capasso, “Flat optics with designer metasurfaces,” Nat. Mater. 13(2), 139–150 (2014).
19. S. Jahani and Z. Jacob, “All-dielectric metamaterials,” Nat. Nanotechnol. 11(1), 23–36 (2016).
20. M. Khorasaninejad, Z. Shi, A. Y. Zhu, W. T. Chen, V. Sanjeev, A. Zaidi, and F. Capasso, “Achromatic metalens
over 60 nm bandwidth in the visible and metalens with reverse chromatic dispersion,” Nano Lett. 17(3), 1819–1824
(2017).
21. R. Pestourie, C. Pérez-Arancibia, Z. Lin, W. Shin, F. Capasso, and S. G. Johnson, “Inverse design of large-area
metasurfaces,” Opt. Express 26(26), 33732–33747 (2018).
22. P. Y. Chen, C. H. Chen, H. Wang, J. H. Tsai, and W. X. Ni, “Synthesis design of artificial magnetic metamaterials
using a genetic algorithm,” Opt. Express 16(17), 12806–12818 (2008).
23. S. Molesky, Z. Lin, A. Y. Piggott, W. Jin, J. Vuckovic, and A. W. Rodriguez, “Inverse design in nanophotonics,” Nat.
Photonics 12(11), 659–670 (2018).
24. M. Mansouree, A. McClung, S. Samudrala, and A. Arbabi, “Large-scale parametrized metasurface design using
adjoint optimization,” ACS Photonics 8(2), 455–463 (2021).
25. Z. Lin, V. Liu, R. Pestourie, and S. G. Johnson, “Topology optimization of freeform large-area metasurfaces,” Opt.
Express 27(11), 15765–15775 (2019).
26. L. Deng and D. Yu, “Deep learning: Methods and applications,” Found. Trends Signal Process. 7(3-4), 197–387
(2014).
27. W. G. Hatcher and W. Yu, “A survey of deep learning: Platforms, applications and emerging research trends,” IEEE
Access 6, 24411–24432 (2018).
28. X. Chen and X. Lin, “Big data deep learning: Challenges and perspectives,” IEEE Access 2, 514–525 (2014).
29. D. Liu, Y. Tan, E. Khoram, and Z. Yu, “Training deep neural networks for the inverse design of nanophotonic
structures,” ACS Photonics 5(4), 1365–1369 (2018).
30. Z. Liu, D. Zhu, S. P. Rodrigues, K.-T. Lee, and W. Cai, “Generative model for the inverse design of metasurfaces,”
Nano Lett. 18(10), 6570–6576 (2018).
31. S. An, C. Fowler, B. Zheng, M. Y. Shalaginov, H. Tang, H. Li, L. Zhou, J. Ding, A. M. Agarwal, C. Rivero-Baleine,
K. A. Richardson, T. Gu, J. Hu, and H. Zhang, “A novel modeling approach for all-dielectric metasurfaces using
deep neural networks,” (2019).
 Research Article Vol. 30, No. 2 / 17 Jan 2022 / Optics Express 3088

32. C. C. Nadell, B. Huang, J. M. Malof, and W. J. Padilla, “Deep learning for accelerated all-dielectric metasurface
design,” Opt. Express 27(20), 27523–27535 (2019).
33. L. Xu, M. Rahmani, Y. Ma, D. A. Smirnova, and K. Z. Kamali, “Enhanced light–matter interactions in dielectric
nanostructures via machine-learning approach,” Adv. Photonics 2(02), 1 (2020).
34. S. Iman, L. Heon, and R. Junsuk, “Double-deep q-learning to increase the efficiency of metasurface holograms,” Sci.
Rep. 9(1), 10899 (2019).
35. S. An, B. Zheng, H. Tang, M. Shalaginov, L. Zhou, H. Li, T. Gu, J. Hu, C. Fowler, and H. Zhang, “Generative
multi-functional meta-atom and metasurface design networks,” ArXiv abs/1908.04851 (2019).
36. A. Sharma, D. Gangwar, B. Kumar Kanaujia, and S. Dwari, “Rcs reduction and gain enhancement of srr inspired
circularly polarized slot antenna using metasurface,” AEU - Int. J. Electron. Commun. 91, 132–142 (2018).
37. Y. Zheng, J. Gao, X. Cao, Z. Yuan, and H. Yang, “Wideband rcs reduction of a microstrip antenna using artificial
magnetic conductor structures,” IEEE Antennas Wirel. Propag. Lett. 14, 1582–1585 (2015).
38. T. Shan, X. Pan, M. Li, S. Xu, and F. Yang, “Coding programmable metasurfaces based on deep learning techniques,”
IEEE J. on Emerg. Sel. Top. Circuits Syst. 10(1), 114–125 (2020).
39. K. Hornik, M. Stinchcombe, and H. White, “Universal approximation of an unknown mapping and its derivatives
using multilayer feedforward networks,” Neural networks 3(5), 551–560 (1990).
40. G. Cybenko, “Approximation by superpositions of a sigmoidal function,” Math. Control. Signals & Syst. 2(4),
303–314 (1989).
41. K. Hornik, M. Stinchcombe, and H. White, “Multilayer feedforward networks are universal approximators,” (1989).
42. S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate
shift,” in International conference on machine learning, (PMLR, 2015), pp. 448–456.
43. C. Doersch, “Tutorial on variational autoencoders,” arXiv preprint arXiv:1606.05908 (2016).