Artificial Neural Networks: A Practical Review of Applications Involving Fractional Calculus

THE EUROPEAN
Eur. Phys. J. Spec. Top. (2022) 231:2059–2095

https://doi.org/10.1140/epjs/s11734-022-00455-3 PHYSICAL JOURNAL
SPECIAL TOPICS
Regular Article
Artificial neural networks: a practical review of

applications involving fractional calculus
E. Viera-Martin1, J. F. Gómez-Aguilar2,a , J. E. Solı́s-Pérez3 , J. A. Hernández-Pérez4 , and
R. F. Escobar-Jiménez1
1
Tecnológico Nacional de México/CENIDET, Interior Internado Palmira S/N, Col. Palmira, C.P. 62490 Cuernavaca,
Morelos, Mexico
2
CONACyT-Tecnológico Nacional de México/CENIDET, Interior Internado Palmira S/N, Col. Palmira, C.P. 62490
Cuernavaca, Morelos, Mexico
3
Escuela Nacional de Estudios Superiores Unidad Juriquilla, Universidad Nacional Autónoma de México, Boulevard
Juriquilla 3001, Juriquilla La Mesa, C.P. 76230 Juriquilla, Querétaro, Mexico
4
Universidad Autónoma del Estado de Morelos/Centro de Investigación en Ingenierı́a y Ciencias Aplicadas, Av. Universidad
No. 1001, Col Chamilpa, C.P. 62209 Cuernavaca, Morelos, Mexico
Received 7 March 2021 / Accepted 13 January 2022 / Published online 12 February 2022
© The Author(s), under exclusive licence to EDP Sciences, Springer-Verlag GmbH Germany, part of
Springer Nature 2022
Abstract In this work, a bibliographic analysis on artificial neural networks (ANNs) using fractional cal-
culus (FC) theory has been developed to summarize the main features and applications of the ANNs.
ANN is a mathematical modeling tool used in several sciences and engineering fields. FC has been mainly
applied on ANNs with three different objectives, such as systems stabilization, systems synchronization,
and parameters training, using optimization algorithms. FC and some control strategies have been sat-
isfactorily employed to attain the synchronization and stabilization of ANNs. To show this fact, in this
manuscript are summarized, the architecture of the systems, the control strategies, and the fractional
derivatives used in each research work, also, the achieved goals are presented. Regarding the parameters
training using optimization algorithms issue, in this manuscript, the systems types, the fractional deriva-
tives involved, and the optimization algorithm employed to train the ANN parameters are also presented.
In most of the works found in the literature where ANNs and FC are involved, the authors focused on
controlling the systems using synchronization and stabilization. Furthermore, recent applications of ANNs
with FC in several fields such as medicine, cryptographic, image processing, robotic are reviewed in detail
in this manuscript. Works with applications, such as chaos analysis, functions approximation, heat transfer
process, periodicity, and dissipativity, also were included. Almost to the end of the paper, several future
research topics arising on ANNs involved with FC are recommended to the researchers community. From
the bibliographic review, we concluded that the Caputo derivative is the most utilized derivative for solv-
ing problems with ANNs because its initial values take the same form as the differential equations of
integer-order.
1 Introduction for describing the dynamical behavior of the ANNs neu-

rons. Hence, in the last decade, many authors employed
Artificial neural networks (ANNs) have emerged as FANN for modeling physics and engineering systems
a promising alternative to simulate systems due to more efficiently and accurately.
their successful applications in several engineering and In this manuscript, we will use the term “ANN
science fields, such as signal processing, image pro- involved with FC” to refer to the FANN. Compared
cessing, control systems, associative memory, to name with the ordinary ANNs, ANNs involved with FC have
a few. Besides, fractional calculus (FC) is an exten- important advantages, such as the description of mem-
sion and generalization of the integer-order calculus, ory and hereditary properties of several processes; and
which its main characteristic is the memory descrip- the system performance is enriched due to one more
tion. When the ANNs are modeled using fractional dif- degree of freedom [1]. Fractional-order systems can
ferential equations (FDE), they are named fractional process information efficiently, improving the simula-
artificial neural networks (FANNs). The FDE is used tions of the integer-order systems, finding more accu-
rate results. Since many of the real-world problems can
a
be generally identified and described by the fractional-
e-mail: [email protected] (corresponding
author)
123
2060 Eur. Phys. J. Spec. Top. (2022) 231:2059–2095
order models [2], it can be expected the same or better analysis of several systems has been investigated since
results could be reached using the FANN. they have been successfully applied in some engineering
The key aspects of this review are the study and fields, such as signal processing, pattern classification,
comprehension of the main derivatives related to FC. control, and optimization [14]. In recent years, FC is
Several definitions of fractional derivatives have been introduced for the stability analysis of nonlinear sys-
proposed, for example, Grünwald–Letnikov, Riemann– tems, allowing us to study the most important stability
Liouville, Caputo, Caputo–Fabrizio, and Atangana– types, such as exponential stability, finite-time stability,
Baleanu fractional derivatives. The Riemann–Liouville uniform stability, global stability, etc. Several control
and Caputo derivatives use the power-law kernel. The techniques have been widely used to guarantee stability
Caputo–Fabrizio derivative uses an exponential ker- on ANN involved with FC, among them, sliding mode
nel, and the Atangana–Baleanu derivative utilizes the control, feedback control, and impulsive control [13,14].
generalized Mittag–Leffler function as the nonsingu- Besides, the ANN involved with FC has been
lar and nonlocal kernel. The equations described by employed in approximation, estimation, control of chaos.
fractional derivatives are highly complex, and there Moreover, it has been found an ANN involved with
exist lots of analytical and numerical methods to solve FC with applications in cryptographic, medicine, sus-
them. These techniques have allowed establishing a tainable energy, images, circuit realization, unmanned
comparison point between the exact solution and the aerial vehicles, and robotics.
approximation carried out by the ANN involved with In the present work, a state of the art review related
FC. Several numerical or analytical methods that have to the ANN involved with FC is carried out. This paper
been developed to solve FDE are, for instance: the is organized as follows: Sect. 2 presents a synthesis
Adams–Bashfort–Moulton method, homotopy pertur- about the FC applied to ANN; in Sect. 3, the ana-
bation method, variational iteration method, Adomian lytical and numerical methods employed to solve the
decomposition method, Laplace transform method, Differential Equations (DE) and FDE that model the
among others [3–10]. concerned systems are reviewed. Subsequently, Sect. 4
The uncertainty of parameters affects the modeling presents a thorough overview of the optimization algo-
and controlling of the systems we are dealing with. rithms employed for the training of ANN involved with
Therefore, parameter estimation of ANN involved with FC; in Sect. 5, the control strategies employed to syn-
FC is crucial for the theoretical study and practi- chronize and stabilize ANN involved with FC are sum-
cal applications [11]. In training ANN involved with marized. Section 6 shows other important applications
FC, the synaptic connection weights between differ- of ANN involved with FC. In Sect. 7, some future direc-
ent neurons are adjusted. The weights training is car- tions about ANN involved with FC are given. Finally,
ried out using optimization algorithms. An optimiza- Sect. 8 presents a summary of the most relevant infor-
tion algorithm is an efficient searching method to solve mation of this research.
constrained optimization problems [12]. The various
optimization algorithms to train ANN involved with
FC include the algorithms based on back-propagation 2 Mathematical preliminaries
(BP), such as gradient descent algorithm (GD) or
Levenberg–Marquardt algorithm (LM) among others, In this section, we present some fractional-order deriva-
and algorithms based on heuristic methods, such as tives widely used in the fractional calculus.
genetic algorithm (GA), simulating annealing algo-
rithm (SA), particle swarm optimization algorithm Definition 1 Let α ∈ R+ and n = α. The fractional
(PSO), and so on. At present, synchronization of operator in the Riemann–Liouville sense is given as fol-
chaotic fractional-order differential systems becomes a lows [17]
challenging and interesting problem due to its poten-
tial applications and the ability to model systems accu-
RL α 1 dn
rately. Amongst all kinds of fractional-order chaos syn- 0 Dt {f (t)} =
Γ(n − α) dtn
chronization, the most commonly employed to synchro- t
nize ANN involved with FC are projective synchroniza- f (τ )
dτ, n − 1 < α ≤ n, (1)
tion, global synchronization, finite-time synchroniza- a (t − τ )α−n+1
tion, quasi synchronization, and adaptive synchroniza-
tion. Nowadays, many authors have published works
where a and t are the limits of operation of RL α
0 Dt {f (t)}
where sufficient conditions are derived, to achieve these
and Γ(·) is the Euler Gamma function.
types of synchronizations on ANN involved with FC
[13,14]. Many control techniques have been used to
Definition 2 Let α ∈ R+ . The fractional operator of
show their synchronization, among them: feedback con-
Grünwald–Letnikov is given by [17]
trol, adaptive control, sliding mode control, impulsive
control, and so on [15]. Moreover, the stability anal-
ysis of ANN is one of the most important and active GL (α
h f ) (x)
Dtα {f (t)} = lim , (2)
areas of research. Consequently, some improved stabi- h→0 hα
lization methods for different types of nonlinear sys-
tems are worthy of further investigation [16]. Stability where,
123
Eur. Phys. J. Spec. Top. (2022) 231:2059–2095 2061
∞

α
k tional dynamics based on various types of FDE, the
(α
h f ) (x) = (−1) f (x − kh) , α > 0, (3)
k obtained results were compared with other numeri-
k=0 cal methods, such as the spectral collocation method,
α meshless method, and reproducing kernel method, to
and k is the generalized binomial coefficient. demonstrate the feasibility of the proposed approach.
The stability of impulsive complex-valued BAM
Definition 3 The Liouville–Caputo fractional deriva- FANN with time-varying delays was studied in Ref.
tive, which will be named as Caputo derivative in the [23]. In this work, the Laplace transform of the Mittag–
rest of this paper, is defined as the convolution of the Leffler function was obtained, and the Mittag–Leffler
local derivative of a given function with a power-law stability of the Caputo fractional derivative was proved.
kernel. Therefore, the derivative of order (α > 0) is For homogeneity reasons, each table presented in
defined as follows [17] this paper shows the ANN architectures that involve:
time delay, time varying-delays, multiple delays, mixed
C α 1
0 Dt {f (t)} = delays, and leakage delay, followed by the architecture
Γ(n − α) of the ANN.
t
f (n) (τ ) Table 1 shows the overview of methods used to solve
× dτ, n − 1 < α ≤ n. (4) DE and FDE applying ANN involved with FC. The
0 (t − τ )α−n+1
types of the differential equation, the referenced work,
the method to solve the DE or FDE, the ANNs archi-
Definition 4 Let f ∈ H 1 (a, b), b > a, α ∈ [0, 1) tectures, as well as the fractional derivative used, are
then, the Atangana–Baleanu fractional derivative in shown in Table 1, as follows:
Liouville–Caputo sense is given as [18] According to the bibliographic analysis developed
in this manuscript, we can affirm that the most used
ABC α AB(α) method to solve FDE and Delayed FDE is the Laplace
0 Dt {f (t)} =
n−α transform method, followed by the Adams–Bashforth–
t
(n) (t − τ )α Moulton method, both of them mainly used the Caputo
× f (τ )Eα −α dτ, n − 1 < α < n, derivative.
0 n−α
where AB(α) is a normalization function. 3.1 Brief analysis about numerical methods
The interest in applying and solving fractional dif-

ferential equations (FDEs) has increased in the last
3 Methods to solve fractional differential decades. Methods such as Laplace/Sumudu pertur-
equations bation [62,67,68,98,114,122,123], Adomian decompo-
sition [46,124,125], homotopy perturbation [126] or
There are different analytical and numerical meth- decomposition have been used to reach this objective.
ods to solve DE and FDE. In this section are intro- However, these methods are faced with the convergency,
duced the methods used to solve the DE and FDE the stability, ability to handle strong non-linearities
on ANN involved with FC. Also, in this section, a [127] even the presence of a persistent memory [128].
brief discussion is carried out to summarize some of For this reason, linear multistep methods are a pow-
these methods. In [19], a chaotic memristive Hopfield erful option to solve fractional differential equations.
FANN was designed using the 4th order Runge–Kutta The Adams–Bashforth method (ABM) is an efficient
numerical method involving the Grünwald–Letnikov numerical scheme that converges toward the exact solu-
derivative. The dynamical properties of the system tion. One can see several types of research whose numer-
were studied. And an adaptive sliding mode control ical results were reached by ABM [3,24, 33,34,42,44,
was used for the system synchronization. In Ref. [20], 45,82,95,97]. It can lead to non-local, non-singular ker-
the problems on the stability and synchronization of nel fractional derivatives [129]. However, ABM requires
quaternion-valued FANN was investigated involving several floating-point operations [128], and this has lim-
Caputo derivative, and the Adams–Bashforth–Moulton itations because of the Lagrange polynomial. Dealing
predictor–corrector method was used to solve the FDE with these limitations, Atangana and Araz [130] devel-
and demonstrate the validity of the theoretical results. oped a numerical scheme based on Newton polynomial,
On the other hand, in [21], an orthogonal FANN was which seems to be accurate.
employed to solve various types of Lane–Emden equa- Therefore, the proposal of numerical schemes for solv-
tions that arise in several physical phenomena. The ing FDEs is still an open field for new accurate propos-
fractional-order Lane–Emden equation was generalized als with the ability to handle strong non-linearities with
by considering its derivative in Caputo sense, the ana- the least computational effort.
lytical approximation to the solution of the FDE was
carried out using Adomian decomposition method. In
Ref. [22], an orthogonal Jacobi FANN was employed
to perform the numerical simulations of nonlinear frac-
123
2062
123
Table 1 Methods to solve fractional differential equations
Differential equation References Method of solution ANN architecture Fractional derivative
DE [24] Adams–Bashforth–Moulton Feed-forward ANN
method
[21] Collocation method Orthogonal FANN Caputo derivative
[21] Reproducing kernel method Orthogonal FANN
[25] Runge–Kutta method ANN
DE [19] Runge–Kutta method Memristive Hopfield FANN Grünwald–Letnikov derivative
FDE [26] ANN
[11, 27–30] FANN
[31] Cellular FANN
[32] Complex-valued FANN
[33, 34] Adams–Bashforth–Moulton Dynamic FANN Caputo derivative
Method
[35–39] Hopfield FANN
[40] Memristive FANN
[20] Quaternion-valued FANN
[41] Recurrent FANN
FDE [42] Adams–Bashforth–Moulton Feed-forward ANN Riemann–Liouville derivative
method
[43] Nonidentical FANN
FDE [44] Adams–Bashforth–Moulton Feed-forward ANN Riemann–Liouville and Caputo
method derivative
FDE [45] Adams–Bashforth–Moulton Recurrent FANN Grünwald–Letnikov derivative
method
FDE [3] Adams–Bashforth–Moulton ANN Atangana–Baleanu in Caputo sense
method derivative
FDE [21] Adomian decomposition Orthogonal FANN Caputo derivative
method
FDE [46] Adomian decomposition Feed-forward ANN Riemann–Liouville and Caputo
method derivative
FDE [46] Chebyshev wavelet method Feed-forward ANN Riemann- Liouville and Caputo
derivative
Table 1 continued
FDE [47–49] Collocation method Feed-forward ANN Caputo derivative
[22] Orthogonal Jacobi FANN
FDE [46] Differential transform Feed-forward ANN Riemann–Liouville and Caputo
derivative
FDE [26] Euler method ANN Caputo derivative
FDE [3] Euler method ANN Atangana–Baleanu in Caputo sense
derivative
FDE [50] Finite difference method Deep FANN Caputo derivative
[22] Jacobi orthogonal FANN

FDE [51] Finite difference method ANN Riemann–Liouville and Caputo
derivatives
FDE [46] Homotopy perturbation Feed-forward ANN Riemann–Liouville and Caputo
method derivative
[44] Feed-forward ANN
[25] ANN
[52] Deep convolutional ANN
[28, 29, 53–58] FANN
[59] BAM FANN
[60] Cohen–Grossberg FANN
[1, 61] Complex-valued FANN
FDE [62] Laplace transform method Fuzzy FANN Caputo derivative
[38] Hopfield FANN
[16, 63] Memristive FANN
[15] Nonidentical FANN
[64] Quaternion-valued FANN
[65] Quaternion-valued
memristive FANN
[41, 66] Recurrent FANN
FDE [67] Laplace transform method Backpropagation ANN Riemann–Liouville derivative
FDE [68] Laplace transform method ANN Grünwald–Letnikov
2063
123
2064
Table 1 continued
123
FDE [26] Legendre method Feed-forward ANN Caputo derivative
[69] FANN
FDE [3] Legendre Method ANN Atangana–Baleanu in Caputo sense
derivative
FDE [47] Shifted Legendre method FANN Caputo derivative
[22] Orthogonal Jacobi FANN
FDE [27] Linear interpolation method Hopfield FANN Caputo derivative
FDE [70] Linear interpolation method Hopfield FANN Grünwald–Letnikov derivative
FDE [22] Meshless method Orthogonal Jacobi FANN Caputo derivative
FDE [71] Mellin transform method FANN Riemann–Liouville, Caputo and
Grünwald–Letnikov
FDE [21] Orthogonal polynomials Orthogonal FANN Caputo derivative
method
FDE [69] Power series expansion FANN Caputo derivative
method
[47] FANN
FDE [72] Power series expansion Cellular FANN Grünwald–Letnikov derivative
FDE [22] Reproducing kernel method Orthogonal Jacobi FANN Caputo derivative
FDE [46] Variational iteration method Feed-forward ANN Riemann–Liouville and Caputo
derivative
[12] FANN
FDE [44] Variational iteration method Feed-forward ANN Riemann–Liouville and Caputo
derivative
[73] Delayed cellular ANN
[74–76] Delayed FANN
[77, 78] Delayed BAM FANN
[79–81] Delayed complex-valued
FANN
Delayed FDE [82] Adams–Bashforth–Moulton Delayed Fuzzy Cellular Caputo derivative
method FANN
[83–89] Delayed Hopfield FANN
[90–92] Delayed memristive FANN
[93] Delayed memristive
quaternion-valued FANN
[94] Delayed quaternion-valued
FANN
Delayed FDE [95] Adams–Bashforth–Moulton Delayed FANN Riemann–Liouville derivative
method
[96] Delayed competitive FANN
Table 1 continued
Delayed FDE [97] Adams–Bashforth–Moulton ANN Atangana–Baleanu in Caputo sense
method derivative
Delayed FDE [98] Chebyshev orthogonal Chebyshev FANN Atangana–Baleanu in Caputo sense
polynomial method derivative
[99] Hopfield FANN
[74, 100–102] Delayed FANN
[103–105] Delayed BAM FANN
[106] Delayed cellular FANN
[107] Delayed Cohen–Grossberg
FANN
[81, 108–112] Delayed complex-valued
FANN
[23] Delayed complex-valued
BAM FANN
[80, 113] Delayed complex-valued
Hopfield FANN
memristive FANN
Delayed FDE [114] Laplace transform method Delayed competitive FANN Caputo derivative
[115] Delayed coupled FANN
[84, 86, 87] Delayed Hopfield FANN
[90, 116–118] Delayed memristive FANN
[119] Delayed memristive BAM
FANN
[120] Delayed memristive
Cohen–Grossberg FANN
[121] Delayed quaternion-valued
FANN
Delayed FDE [98] Laplace transform method Chebyshev FANN Atangana–Baleanu in Caputo sense
derivative
Delayed FDE [98] Shifted Legendre method Chebyshev FANN Atangana–Baleanu in Caputo sense
derivative
2065
123
4 Optimization algorithms for training Also, Chen et al. [135] implemented an adaptive
artificial neural networks fractional-order BP ANN. This technique uses the pop-
ulation extremal optimization, as well as the fractional-
order GD training algorithms. The method was devel-
This section presents the works found in the review oped to solve handwritten digit recognition problems.
of the state of the art where optimization algorithms Population extremal optimization algorithms were used
are employed for the training of ANN involved with to optimize the initial connection weight parameters,
FC. First, the works with a fractional approach are and the fractional GD was used to update these con-
described. Afterwards, the proposals under integer- nection weight parameters.
order operators are described. Finally, a FANN was proposed in Ref. [136] for the
identification of three different systems. The FANN was
Optimization algorithms with fractional approach trained by using the GD algorithm and the Grünwald–
Letnikov derivative. In this work, the three systems
In literature, there were found six works where the that we identified are two benchmark systems and one
fractional GD algorithm was implemented, four works experimental system. The benchmark systems are a
based on fractional BP algorithm, and one research hairdryer, consisting of a mesh of resistor wires heat-
where a Darwinian particle swarm optimization algo- ing the air at the entrance of a pipeline, and a hystere-
rithm of fractional-order was developed, all of them sis model consisting of the Bouc–Wen model used to
were used for the training of ANN involved with FC. represent hysteresis effects in mechanical engineering.
Next, these works are discussed. Besides, the experimental acoustic duct system is based
on an acoustic waves pipeline. The results demonstrated
Gradient descent algorithm (GD) that the fractional gradient descent algorithm allowed
accurate estimations with a reduced number of parame-
In Ref. [131], the author implemented a fractional GD ters, compared with other works found in the literature
algorithm to derive the fractional back-propagation where the gradient descent algorithm of integer-order
through time (FBPTT) algorithm for recurrent ANN, was employed.
based on the Riemann–Liouville derivative; this frac-
tional algorithm was able to solve three estimation
problems, namely: nonlinear system identification, clas- Backpropagation algorithm (BP)
sification of pattern and Mackey–Glass chaotic time
series prediction, outperforming the conventional back- In Ref. [137], the authors developed a Hopfield FANN
propagation through time performance. in the form of an analog circuit. To carried out this,
Other interesting works were proposed in Refs. they used factorial, as well as steepest descent fractional
[132,133]. They got a fractional GD back-propagation approaches.
method based on the Caputo derivative for training The authors in Ref. [138] developed a fractional-order
an ANN [132] and deep BP ANN [133]. They derived BP ANN for improving the performance of the ordi-
the error function monotonicity, the proposed algo- nary first-order BP ANN, which was trained by an
rithms presented weak convergence, and numerical sim- improved fractional-order steepest descent method. The
ulations demonstrated the competitive performance of proposed approach showed to be capable of finding the
the presented fractional models. Compared with classi- global optimal solutions. The BPFANN was compared
cal integer-order models, the fractional models showed with a classic first-order BPANN by means of an exam-
significant advantages of memory storage and heredi- ple function approximation, fractional-order multi-scale
tary characteristics. The authors in Ref. [133] carried global optimization, and two comparative performances
out a comparison between two different methods to with real data, involving Grünwald–Letnikov deriva-
test their performances, the first was an ordinary BP tive. The BPFANN was compared with a classic first-
ANN, and the second one was a fractional-order deep order BPANN. To carry out the methods comparison,
BP ANN. The results of the training and testing showed they were used an approximation function, a fractional-
that fractional-order deep BP ANN has a better per- order multi-scale global optimization method, and two
formance. real data sets. The FANN was developed involving
Afterward, as the fractional-order gradient could the Grünwald–Letnikov derivative. The BPFANN was
not converge to the real extreme point. In Ref. [134], superior to the classic first-order BPANN in terms of
the authors designed a new fractional-order gradient finding the global optimal solution.
method based on Caputo derivative for the BP of con- Additionally, in Ref. [139], the authors implemented
volutional ANN. In this work, the parameters within a fractional observer ANN for high complex fractional-
the layers were updated using the fractional gradient order nonlinear systems, involving the Caputo deriva-
method, but propagations between layers used integer tive, for estimating the state variables of a fractional-
gradients to keep the chain rule. In fact, the proposed order nonlinear chaotic system with the unknown
fractional-order gradient method guaranteed the con- dynamic model. A new fractional error back-propagation
vergence to a real extreme point, fast convergence, high learning algorithm was developed to adapt the weights
accuracy, and ability to escape local optimal point in of the ANN. This method could eliminate the effect of
ANN when compared with integer-order ANN. uncertainties and unmodeled dynamics of the system,
123
showing a fast convergence. The fractional observer was Gradient descent algorithm (GD)
better than other observers of integer-order in terms of
accuracies for fractional systems, distinguishing exter- Firstly, in Ref. [67], the authors proposed a fractional
nal disturbances, and modeling uncertainties more effi- PID controller with self-tuning parameters based on
ciently. BP ANN. The discretization method and the design
method of the controller were discussed. The authors
used the Riemann–Liouville derivative to develop the
fractional controller. The controller parameters were
tuned by an ANN, which was optimized by the GD
Darwinian particle swarm optimization algorithm
optimization algorithm. The fractional PID controller
(FO-PSO) was more flexible than the ordinary. The fractional con-
troller allowed the authors to adjust better its dynam-
In Ref. [24], the authors developed a feed-forward ANN ical properties than the ordinary PID. Moreover, the
optimized by applying a fractional-order Darwinian fractional PID controller with a self-tuning parameter
particle swarm optimization algorithm (FO-DPSO) to based on a BP ANN kept features of a normal fractional
calculate better solutions to the nonlinear second-order PID controller. It has better flexibility because an ANN
ordinary differential equations representing the corneal was introduced to develop the self-tuning controller.
shape model (CSM). The authors used the Grünwald– Subsequently, a set of fractional differential equations
Letnikov derivative. Adams’s numerical solver was used of initial value problems constructed from cosine basis
as the reference solution. PSO-DPSO was compared functions with adjustable parameters were solved in
with the hybridization between the PSO algorithm Ref. [140] using an ANN and the Caputo derivative.
and the Active set algorithm (PSO-ASA). The result Numerical solutions were obtained for a single FDE, as
showed that ANN-based FO-DPSO was more accurate well as for systems of coupled FDE. The numerical solu-
in the solutions with fewer residual errors. Performance tions were obtained by training the ANN repeatedly by
matrices like MAD, TIC, and ENSE were used to test means of the GD algorithm. Numerical results were in
the efficiency of the proposed approach, demonstrating good agreement with the exact solutions of the FDE.
that the proposed methodology was better in terms of Similarly, Jafarian in Ref. [141] employed the Caputo
less number of function evaluations, mean time value, derivative over a bounded domain to approximate series
ENSE, TIC, MAD. FO-DPSO was an excellent tech- solutions of a class of initial value FDE. The ANN was
nique for tuning the unknown weights involved in the used to solve a fractional type ordinary DE. The orig-
solution designed with ANNs. inal differential equation was transformed into a mini-
mization problem, which was solved using an accurate
ANN model for computing the parameters accurately.
The authors achieved this using the GD procedure for
Optimization algorithms with classical approach training a feed-forward ANN. The proposed algorithm
was an efficient tool for finding the unknown series coef-
From the bibliographic review, there were found seven ficients. The obtained simulation results were compared
works where the classical GD has been used as an with the exact solutions, Chebyshev wavelet method,
optimization algorithm for training the ANN involved and the Homotopy perturbation method, reported in
with FC. In two research works, the classical BP the literature. Finally, the authors demonstrated that
algorithm was proposed. Moreover, in six works, the numerical simulations carried out by the ANN trained
classical Levenberg–Marquardt algorithm (LM) was using the GD algorithm were similar to the solutions
implemented for training the ANNs. On the other reported in the literature.
hand, some other classical optimization algorithms have On the other hand, a Chebyshev functional link ANN
been employed with this purpose, such as interior was proposed by Kheyrinataj and Nazemi in Ref. [98]
point algorithm (IPA), genetic algorithm hybridized to model linear and nonlinear delay fractional optimal
with pattern search algorithm (GA-PS), sequential control problems involving Atangana–Baleanu deriva-
quadratic programming algorithm (SQP), Broyden– tive. The trial solutions were approximated by applying
Fletcher–Goldfarb–Shanno algorithm (BFGS), chaotic the Chebyshev functional link ANN, which was trained
differential evolution algorithm (CDE), simulated using the GD algorithm. This work presented the first
annealing algorithm (SA), particle swarm optimiza- application of delay fractional optimal control with
tion algorithm (PSO), (SA) hybrid with particle swarm mixed control-state constraints employing a fractional-
optimization algorithm (PSO) obtaining the (PSO- order derivative with the nonlocal and nonsingular ker-
SA) algorithm, particle swarm optimization algorithm nel using the Chebysehv functional link ANN approach.
hybrid with enhanced fruit fly algorithm (PSO-EFF), Besides, Antil in Ref. [50] proposed a deep FANN for
particle swarm optimization algorithm hybrid with arti- the time-discretization of a fractional in time nonlinear
ficial bee colony algorithm (PSO-ABC), and stochastic ordinary DE, employing Caputo derivative. The frac-
inertia weight particle swarm optimization (SIWPSO) tional ordinary DE was minimized by the learning algo-
algorithm. rithm, solving several issues, such as network instability,
All the above-mentioned works are going to be vanishing, exploding gradients, long training times, and
explained in detail as follows: inability to approximate non-smooth functions. Keep-
123
ing track of history in this manner improved the vanish- robust control for applications in unmanned aerial vehi-
ing gradient problem and could potentially strengthen cles (UAVs). The ANN was trained to provide the coef-
feature propagation. It was numerically illustrated the ficients of a finite impulse response (FIR) of approxima-
improvement in the vanishing gradient issue using of the tion type. Some trajectories were described properly by
proposed deep FANN allowing a better learning ability. the FIR controller, and the feed-forward ANN, which
The deep FANN was better capable of passing informa- was trained by the LM algorithm.
tion across the network layers maintaining the relative Next, in Ref. [145] the authors developed a new
gradient magnitude across the layers, compared to the method to detect unilateral hearing loss (left-sided and
standard deep ANN and standard Residual ANN. The right-sided); the fractional Fourier transform (FRFT)
deep FANN improved the vanishing gradient issue due was employed to detect hearing loss more efficiently
to the memory effect, and it handled much better the and accurately. The classifier was a feed-forward ANN
nonsmooth data due to the network’s ability to approx- trained by the LM algorithm. Some magnetic resonance
imate non-smooth functions. images were obtained from studies with real patients.
Finally, the authors in Ref. [49] tested the abil- The combination of fractional Fourier transform, the
ity of a perceptron ANN to approximate functions to principal component analysis, and the neural network
solve fractional infinite-horizon optimal control prob- as the classifier, showed accuracies higher than 95%
lems involving Riemann–Liouville and Caputo deriva- concerning the experimental data [145].
tives. The authors used the GD algorithm for training On the other hand, Zúñiga-Aguilar [3,97] used an
the ANN. There were no other reported works about ANN to get the approximated solution of fractional
solving this problem in the literature. Numerical sim- differential equations of Atangana–Baleanu type in
ulations demonstrated the feasibility and efficiency of Caputo sense with delay and without delay, respec-
using the proposed method for solving optimal control tively. In both cases, the network’s parameters opti-
problems. mization was carried out using the LM algorithm. The
results of both ANNs were compared with the analyt-
ical solutions and the numerical simulations obtained
Backpropagation algorithm (BP) through the Adams–Bashforth–Moulton method. Dif-
ferent performance indices were calculated to show the
The BP algorithm was used for training a Master-Slave effectiveness of the ANNs. The ANN’s were able to
FANN based on Caputo derivative in Ref. [27]. The achieve approximate solutions with good precision and
master network was composed of two Hopfield net- fast convergence.
works, meanwhile and the slave network was a BP net- Subsequently, Kadam et al. developed the artifi-
work, doing the BP the system error. The Master-Slave cial ANN approximation of fractional derivative opera-
FANN showed to have the highest asymptotic conver- tors such as Grünwald–Letnikov and Caputo fractional
gence rate and the smallest system error compared with derivatives. LM algorithm was used for training the
Master-Slave ANN of integer-order. ANN, considering the mean squared error between the
Moreover, in Ref. [142] the authors found the numeri- outputs of derivatives and the approximations for val-
cal solution of FDE by employing the Chebyshev ANN, idation. Thus, the approximations were computation-
Riemann–Liouville, and Caputo derivatives. BP algo- ally fast when compared with the numerical evaluation
rithm was used to train the feed-forward Chebyshev of fractional-order derivatives.
ANN. The accuracy of the proposed method was shown In a recent paper [21], the authors showed the design
by comparing the analytical solutions with the numer- of a single layer orthogonal ANN for approximating the
ical results. The obtained results showed a good agree- solutions of different types of Lane–Emden equations in
ment with analytical solutions. The comparison results the Caputo sense. The fractional-order Legendre func-
showed that the Chebyshev ANN is a capable tool for tions in Caputo sense were used as the hidden layer
solving linear and nonlinear problems.
activation function, while the LM algorithm was used
Nouh [143] modeled the fractional polytropic gas
to train the ANN. The obtained results were compared
spheres, which have several applications in physics,
with some other numerical methods and with the exact
astrophysics, engineering, and so on. Thus, the frac-
solution, showing that the proposed orthogonal ANN
tional Lane–Emden differential equations of the frac-
was accurate and feasible.
tional polytropic gas spheres phenomena were solved
Finally, Hadian Rasanan in Ref. [22] implemented
employing ANN-based on back-propagation training
a fractional ANN. The authors used fractional-order
algorithm, reaching the training of the ANN with small
Jacobi functions as the activation function of the hidden
errors predicting the values of fractional Lane–Emden
layer. And the identity function was used as the activa-
functions.
tion function of the output layer. The goal of this work
was to approximate the solution of FDE and partial
Levenberg–Marquardt algorithm (LM) FDE involving Caputo derivative. LM was the train-
ing algorithm employed. Thereby, the proposed ANN
Firstly, Efe and Member in Ref. [144] developed an ana- had the ability to reach high accuracy with few neu-
log PID controller using an approach of feed-forward rons. The effectiveness of the proposed ANN was vali-
ANN. The implementation of fractional-order operators dated applying linear and nonlinear fractional dynam-
in the PID controller was discussed for establishing a ics. The numerical results were compared with the
123
results obtained from other ANN and some numerical ing the Riemann–Liouville and the Caputo derivatives.
experiments, demonstrating that the proposed model is The obtained results with the proposed methodology
accurate, fast, and feasible. coincided with the exact solution based on the Adams–
Bashforth–Moulton technique. Even, the results were
Interior point algorithm (IPA) more accurate than the obtained with both the mod-
ified homotopy perturbation and the enhanced Homo-
Regarding the use of the IPA algorithm for the train- topy perturbation methods showing the effectiveness of
ing of ANN involved with FC, Asif in Ref. [12] found the proposed scheme.
the solution of fractional systems governed by the
initial value problems of the Bagley–Torvik equation Broyden–Fletcher–Goldfarb–Shanno algorithm
employing a FANN trained with the IPA algorithm (BFGS)
via Caputo and Riemann–Liouville derivatives. The
designed method was evaluated on different initial value An adaptive fractional-order PID controller using ANN
problems of the equation. A comparison between the was designed based on auto-tuning neurons, involving
proposed method and several available criteria, such as Caputo derivative in [51]. The Nelder–Mead simplex
an exact solution; Podlubny numerical techniques; an search method and BFGS algorithm were used for the
analytical solver based on variational iteration method; parameters tuning. The proposed controller was more
and a reported solution of stochastic solvers based on robust in comparison with conventional controllers [51].
hybrid approaches, allowed to verify the effectiveness of Also, the BFGS quasi-Newton algorithm was
the designed method. Concluding, this efficient compu- employed for training a perceptron ANN involving
tational technique based on FANN, optimized with IPA, Caputo derivative in Ref. [26] and [148] to solve FDE
was able to find the solutions of different variants of and approximate the solution of a fractional optimal
Bagley–Torvik equations in a more accurate way than control problem, respectively. In [26], the authors vali-
other stochastic techniques. dated their method by solving different types of multi-
Also, in Ref. [146] authors found the approximate term FEEs. In Ref. [148], the authors validated their
solutions of nonlinear quadratic systems based on Ric- proposal by comparing their results with other investi-
cati equations of fractional-order by means of FANN gations found in the literature.
trained with IPA algorithm. The obtained results were
compared with the exact solutions proving the effective- Chaotic differential evolution algorithm (CDE)
ness of the proposal. This method matched more closely
with the standard solution obtained from Adams– In Ref. [149], the authors implemented a wind turbine
Bashforth–Moulton method than the modified homo- pitch control for regulating the speed of the rotor and
topy perturbation method and enhanced homotopy per- power production. The authors proposed a fractional-
turbation method. The average time consumed by the order PID combined with a radial basis function ANN
IPA for a run was lower that than other stochastic tech- to improve the performance and alleviating the mechan-
niques based on the PSO and GA algorithms. ical loads. The ANN was trained with the CDE algo-
rithm. The fractional-order PID presented better per-
Genetic algorithm and pattern search (GA-PS) formance and robustness when comparing with other
controllers. Moreover, the FOPID controller alleviated
A new method to train ANN involved with FC was mechanical loads in a better way, compared with other
developed in [147], where a fractional-order system control techniques, such as the PID, PI, radial basis
represented by Bagley–Torvik equation was solved by function PI, and fractional-order PI controller.
means of feed-forward ANN. This new method is based
on evolutionary computational, and it is called the GA Simulated annealing algorithm (SA)
algorithm hybrid with the PS technique. Besides, in this
work, the Riemann–Liouville derivative was used. The The chaotic behavior of fractional-order Chua’s system
proposed method was successfully applied to different was studied by means of fractional Laplace transform,
forms of the equation, and the results were compared the activation function for the ANN was the Mexican
with a standard approximate analytic solution, stochas- hat wavelet. The ANN was trained using SA and the
tic numerical solvers, and exact solutions. The GA, PS, analytical solution for the system could be approxi-
and Ga hybrid with PS (GA-PS) optimizer algorithms mated. The accurate approximated solutions, the phase
were compared against each other for evaluating the plots of the Lyapunov exponent spectrum, and bifur-
performances of the training algorithms, obtaining the cation maps of the dynamical evolution of fractional
best results with the GA-PS algorithm. Chua’s system were achieved. Mexican Hat Wavelet-
based ANN with SA and fifth-fourth Runge–Kutta
Sequential quadratic programming algorithm (SQP) method, were proposed to attain the solutions of frac-
tional Chua’s model. Using Caputo derivative, cubic
In Ref. [44], the authors applied a feed-forward ANN nonlinear was solved efficiently and the Chua’s circuit
and SQP algorithm for the training of weights to obtain variables were optimized for different fractional values,
the solution of nonlinear quadratic Riccati FDE involv- in [25].
123
Subsequently, in Ref. [150] the authors found the cal method, the stochastic numerical solvers, and exact
numerical solution of delayed FDE based on the appli- solutions. The weights training was proved with other
cation of neural minimization using Chebyshev simu- stochastic algorithms, such as SA, GA, GA hybridized
lated annealing ANN and Legendre simulated anneal- with SA (GA-SA), PSO, and PSO hybridized with SA
ing ANN. Chebyshev and Legendre polynomials were (PSO-SA). The best optimization results were obtained
used with SA to reduce mean square error and get by PSO-SA algorithm.
more accurate numerical approximations. This work In Ref. [42], the PSO-SA algorithm was employed
was based on the functional link ANN with optimiza- for the training of a feed-forward ANN. In this case,
tion through thermal minimization. Caputo’s definition the ANN approximated the mathematical model of
was employed for calculating the fractional derivative FDE using Riemann–Liouville and Grünwald–Letnikov
in the subsequent procedure, and the learning method- derivatives. Comparison between the obtained results
ology used in this work was the SA algorithm. The and the available exact solutions, analytic solutions,
obtained results were validated using various experi- and standard numerical techniques (including both
ments, numerically. And graphically with error analy- deterministic and stochastic approaches) was carried
sis to demonstrate the accuracy and efficiency of the out, showing that This approach was properly employed
proposed approach. to solve different problems associated with linear and
nonlinear ordinary FDE. The training of weights was
implemented using PSO, GA, GA-SA, and SA algo-
Particle swarm optimization (PSO)
rithms, but the best results were obtained using the
PSO-SA algorithm.
In Ref. [34], the authors presented a fractional-order
dynamic ANN trained by the PSO algorithm for iden-
tifying the Damavand tokamak plasma behavior using Particle swarm optimization and enhanced fruit fly
the Caputo derivative. The system stability was demon- (PSO-EFF)
strated based on the Lyapunov-like analysis. The per-
formance of the proposed approach was compared In Ref. [68], authors developed a nonlinear neural
with experimental data and the integer-order ANN fractional-order proportional integral derivative con-
approach. The comparison results showed that the troller based on ANN and involving the Grunwald–
fractional-order dynamic ANN was higher accurate Letnikov derivative, applied to the motion control of
than the dynamic ANN of integer-order. a nonholonomic differential drive of a mobile robot.
Zhang and Yang in Ref. [151], studied the optimal The hybridization of a modified adaptive PSO and the
quasi-synchronization problem for delayed memristive EFF optimization algorithms were used for tuning the
FANN involving Caputo derivative. Fractional-order parameters of the fractional-order PID controller based
inequalities and aperiodically intermittent controllers on ANN. The fractional-order PID controller decreased
were proposed to guarantee the quasi-synchronization the control signals that drive differential drive mobile
of the system. Mittag–Leffler function allowed to get robot motors by approximately 45% compared with the
the stability result of the fractional-order delayed sys- PID based on ANN of integer-order, and thus reduced
tem. Synchronization of the delayed memrisitve delayed the energy consumption in circular trajectories. Numer-
FANN was ensured thanks to matrix inequalities. The ical simulations demonstrated that the performance of
control parameters were optimized, and the smaller the designed fractional controller was excellent com-
control energy was obtained applying the PSO algo- pared with nonlinear controllers of integer-order on the
rithm. Simulation examples showed the correctness of trajectory tracking of the differential drive mobile robot
the proposed method. with different trajectories as study cases.
Aslipour and Yazdizadehin in Ref. [34] optimized
a dynamic FANN involving Caputo derivative. The
Particle swarm optimization and artificial bee colony
authors used the PSO algorithm to identify the behav-
ior of a wind turbine in operation. The results obtained (PSO-ABC)
from dynamic FANN were compared with the results
obtained from dynamic ANN of integer-order, high- In the current year, Mohammadzadeh and Kayacan in
lighting that the fractional method was more accurate. [152] developed an adaptive fractional-order fuzzy con-
trol method in the sense of Caputo’s definition to con-
trol the frequency in an AC microgrid (MG). A sequen-
Particle swarm optimization-simulated annealing tial general type-2 fuzzy system based on the radial
algorithm (PSO-SA) basis ANN was presented for online modeling of the fre-
quency response of the MG. The parameters were opti-
In Ref. [46], the authors developed a feed-forward ANN mized using of a hybridized approach between PSO and
used for approximating the solution of nonlinear Riccati ABC algorithms. The learning algorithm was examined
FDE using Riemann–Liouville and Caputo derivatives. using white noise as the control input. It was demon-
The network training was carried out using the particle strated that the proposed identification scheme results
swarm optimization algorithm (PSO) hybridized with were in good performance even in a noisy environment.
the simulated annealing algorithm (SA). The results The results had robust performance in the presence of
were compared with the standard approximate analyti- variation of solar radiation, wind speed, load distur-
123
bance, and time-varying dynamics of the other units On the other hand, sliding mode control was employed
of MG. The proposed control approach was compared in [15,58,162–166] to reach: projective, global pro-
with the conventional PI controller, classic fuzzy, and jective, mixed projective synchronization, and syn-
PSO-fuzzy PI controllers. The results showed that the chronization using the Caputo and Riemann–Liouville
proposed scheme has better performance. derivatives.
Also, feedback control was found in several works
on ANN involving FC. This control strategy was
Stochastic inertia weight particle swarm applied in Ref. [20,43,53,56,59,61,90–92,96,116–119,
optimization (SIWPSO) 154,167–183], and aiding to achieve: hybrid projec-
tive, finite-time projective, projective, quasi-projective,
Optimization of synchronization for delayed memris- quasi-uniform synchronization, stability, finite-time sta-
tive FANN was investigated in Ref. [153], involving bility, synchronization, and global stability using
Caputo derivative. It was designed an appropriate con- Caputo and Riemann–Liouville derivatives.
troller where the drive system was able to synchro- Other control strategies, such as adaptive control,
nize with the response system. Synchronization condi- impulsive control, or washout filter control were
tions were obtained thanks to the linear matrix inequal- employed in [101,184–187] to attain: projective, adap-
ity, along with fractional-order Lyapunov methods. The tive, global, chaotic synchronization, and global sta-
FANN was trained using of the SIWPSO algorithm, bility, using the Caputo derivative and Grünwald–
the target function was the minimal sum of control Letnikov derivatives.
energy expressed by the integral square error-index, The most used fractional derivative to guarantee the
where the Riemann–Liouville derivative was used to synchronization of these types of systems is the Caputo
get the approximated value of the target function. It derivative, followed by the Riemann–Liouville and the
was obtained a better controller with low control energy Grünwald–Letnikov derivatives, respectively. We found
and integral square error (ISE) index. The optimal con- that the Grünwald–Letnikov derivative was only used
trol parameters of the proposed model were obtained in one work [187].
by computing the SIWPSO algorithm, which was an The information given in the above sections is sum-
improved intelligent algorithm. A simulation was pro- marized in Tables 2 and 3. The tables show the rela-
vided to demonstrate the feasibility of the proposed tionship among the control strategy, ANN architecture,
results. the fractional derivative, and the goal achieved (appli-
cation), as follows:
5 Synchronization and stabilization of ANN

involved with FC 5.2 ANN involved with FC to reach stability
The FC applied to ANNs allows simulating systems Several authors in literature stabilized ANN involved
more accurately than by the classical approach, due to with FC, thanks to the FDE employed in these sys-
the FC properties as the nonlocality and the memory tems. In this sense, different types of fractional deriva-
description of FC. From the bibliographic review, we tives have been used to reach systems stabilization. Fol-
can affirm that the main applications of FC applied to lowing, we will present some related works to systems
ANN are system stabilization, systems synchronization, stabilization using ANN involved with FC.
and the training of these systems through optimization Caputo derivative was employed to reach the expo-
algorithms. nential stability, in Refs. [32,171,232], the uniform sta-
This section is organized as follows: first, the control bility in [76,79,233–242], the finite-time stability in
strategies carried out to guarantee the synchronization Refs. [75,85,103,175,243–248], the stability and bifur-
and stabilization of ANN involved with FC are summa- cation in Refs. [36,37,41,73,74,77,80,108,249–251], and
rized. Second, other examples of stabilization reached the quasi-uniform stability, fractional input stability,
using FC on ANN are overviewed. global stability, and global periodicity in Refs. [252–
254], respectively.
Similarly, the Caputo and Riemann–Liouville deriva-
5.1 Control strategies employed to synchronize and tives were employed to reach: the global stability in
stabilize ANN involved with FC Refs. [16,28,38,59,62,86,95,117,174,180,185,255–259],
the stability in Refs. [35,41, 54,57,66,84,87,113,155,
Several authors have used control strategies to syn- 260–268].
chronize and stabilize different architectures of ANN Also, the Riemann–Liouville derivative was used to
involved with FC. A brief overview of them will be reach synchronization stability in Refs. [252–254,269,
described as follows: 270].
On the one hand, in [11,13,55,154–161] authors Concluding, the Caputo derivative is the most used
employed adaptive control to reach: projective, mixed fractional derivative for investigating the stability of
projective synchronization, and synchronization using the ANN involved with FC, followed by the Riemann–
the Caputo derivative. Liouville derivative. Nowadays, Caputo and Riemann–
123
Table 2 Control strategies employed to synchronize and stabilize ANN involved with FC
Control References ANN architecture Fractional derivative Application
strategy
Adaptive [188] ANN Synchronization
control
[156] Chebyshev ANN Synchronization
[161] Fuzzy ANN Synchronization
[13, 160] FANN Global projective
synchronization
[11, 154] FANN Global synchronization
[55, 155] FANN Synchronization
[189] Chaotic FANN Approximation and
stability
[82] Fuzzy cellular FANN Caputo derivative Global stability and
synchronization
[159] Memristive FANN Projective
synchronization
[190] Memristive FANN with Synchronization
leakage delay
[191] Radial basis function Synchronization
FANN
[158] Delayed FANN Mixed projective
synchronization
[192] Delayed BAM FANN Stability and
synchronization
[157] Delayed fuzzy FANN Projective
synchronization
[109] Delayed Quasi-projective and
complex-valued complete
FANN synchronization
[193] Delayed competitive Projective
FANN Synchronization
Adaptive [194] Hopfield FANN Caputo derivative Stability
sliding mode
control
[195] Delayed fuzzy FANN Projective
synchronization
[52] Deep convolutional Robotic manipulators
ANN
[196] Deep recurrent ANN Synchronization
[165] Radial basis function Stability
ANN
[197] Radial basis function HIV infection model
ANN
[198, 199] Recurrent ANN Approximation
[198] Recurrent ANN Stability
Sliding Mode [58] FANN Caputo derivative Synchronization
Control
[200] Chaotic FANN Syncrhonization
[201] Memristive MAM Fixed-time
[15] Nonidentical FANN Projective
synchronization
[166] Delayed Hopfield Adaptive
[162] Delayed nonidentical Projective
[164] Hopfield FANN Finite-time stability
123
Table 2 continued
strategy
Sliding Mode [163] Nonidentical FANN Riemann–Liouville Global projective
Control Derivative synchronization
Sliding Mode [19] Memristive Hopfield Grünwald–Letnikov Synchronization
Control FANN derivative
Pinning [202] Delayed BAM FANN Caputo Derivative Quasi-pinning
Control synchronization and
stability
[107] Delayed Stability and pinning
Cohen–Grossberg synchronization
FANN
Feedback [40] Memristive FANN Caputo derivative Quasi-synchronization
pinning
control
[56, 171] FANN Stability and
synchronization
[203] FANN Robust finite-time cost
control
[29] FANN Hybrid projective
synchronization
[181] FANN Global synchronization
[204] Delayed FANN Global synchronization
[173] Delayed FANN Synchronization
[90] Delayed FANN Hybrid projective
synchronization
[205] Delayed FANN Projective
synchronization
[59] BAM FANN Stability
[206] BAM FANN Global Stability
[168] Delayed BAM FANN Synchronization
[78] Delayed BAM FANN Global synchronization
[207] Delayed BAM FANN Finite-time stability
[78] Delayed BAM FANN Global synchronization
[53] Chaotic FANN Synchronization
[172] Memristive FANN Synchronization
[118, 208] Delayed memristive Synchronization
FANN
[177] Delayed memristive Finite-time projective
[209] Delayed memristive Projective
[117, 174] Delayed memristive Stability and
[210] Delayed memristive Global stability
FANN
[211] Delayed memristive Synchronization
FANN
Feedback [116] Delayed memristive Caputo derivative Finite-time
Control FANN synchronization
[92] Delayed memristive Quasi-uniform
[91] Delayed memristive Quasi-synchronization
FANN
123
Table 2 continued
strategy
[180] Delayed memristive Global synchronization
FANN
[212] Delayed memristive Asymptotic stability
FANN
[119] Delayed memristive Finite-time
BAM FANN synchronization
[182] Memristive recurrent Finite-time
[93] Delayed memrisitve Synchronization and
quaternion-valued stability
FANN
[175] Cohen–Grossberg Finite-time stability
memristive FANN and synchronization
[213] Quaternion-valued Finite-time
[214] Quaternion-valued Synchronization
BAM FANN
[215] Quaternion-valued Global stability
Memristive FANN
[20] Quaternion-valued Global synchronization
FANN and global stability
[216] Delayed Global synchronization
quaternion-valued
FANN
[176] Fuzzy cellular Finite-time stability
memristive FANN and synchronization
[217] Fuzzy FANN Asymptotic stability
[218] Coupled discontinuous Finite-time
[219] Delayed Synchronization
non-autonomous
FANN
[114] Delayed competitive Global asymptotic
FANN stability
[115] Delayed coupled FANN Robust asymptotic
with synchronization
[179] Radial basis function Projective
ANN synchronization
[61] Complex-valued Quasi-projective
recurrent FANN synchronization
[109] Delayed Quasi-projective and
complex-valued complete
[167, 183] Delayed Synchronization
complex-valued
FANN
[170] FANN Synchronization
[220] Delayed FANN Stability and
synchronization
123
Table 2 continued
strategy
Feedback [96] Delayed FANN Riemann–Liouville Synchronization
Control derivative
[43] Nonidentical FANN Global synchronization
[178] Memristive FANN Projective
synchronization
[222] Complex-valued Global asymptotic
memristive FANN synchronization
[223] Delayed memristive Global projective
BAM FANN synchronization
Liouville derivatives are the only two derivatives used Liu and Fei in Ref. [165], approximated the nonlinear
to reach stability in these systems. Dual Radial Basis functions ANN and the upper bound
Table 4 summarize research works related to ANN of estimated disturbances, improving the system sta-
involved with FC to reach stability. bility and robustness, involving Caputo derivative. The
From the four tables shown in this manuscript: it’s ANNs weights were updated online to approximate the
important to mention that temporal models with recur- dual Radial Basis functions ANN structures, applied to
rent FANN have been developed in works: [41,45,79, a control system.
102,182,198,199,263,266]. In Ref. [189], an adaptive control based on ANN
was used to approximate unknown nonlinear functions
using the fractional Lyapunov stability criterion and
the backstepping technique, to control an uncertain
6 Other interesting applications of ANN fractional-order Chua–Hartley chaotic systems. The
involved with FC ANN was employed to approximate unknown system
uncertainties and external disturbances. The numerical
From the bibliographic analysis, there were found works simulation was given to demonstrate the effectiveness
where ANN was involved with FC to reach goals, such of the proposed approach.
as approximation of functions, description of chaos, On the other hand, Lu and Wang in Ref. [288] devel-
estimation, global dissipativity, periodicity, and mod- oped the adaptive ANN tracking control using back-
eling heat transfer process. Other researches have been stepping technology for the fractional-order chaotic per-
applied to the different areas of science and engineer- manent magnet synchronous motor with the immea-
ing, such as medicine, image encryption, robotic, among surable state, parameter uncertainties, and external
others. The most important works related to these load disturbance, involving Caputo derivative. The pro-
applications will be described in detail in this section, posed approach employed a Chebyshev ANN and a
as follows. state observer to approximate the unknown functions
and estimate the unmeasurable state. In this work, the
simulation results were presented to demonstrate the
ANN involved with FC: applications in correctness of the proposed methodology.
approximation of functions In Ref. [289], authors developed an adaptive ANN
control based on command filtered backstepping method
In several research works, there were approximated for fractional-order permanent magnet synchronous
functions with the aid of ANN involved with FC. Some motor with parameter uncertainties and unknown time
of these works will be explained in detail below: delays, involving Caputo derivative. The unknown
The fractional quantitative approximation of real- parameters, as well as the load disturbance, were
valued functions involving Caputo derivative was car- approximated by using ANN. The time delays uncer-
ried out on feed-forward ANN in Ref. [286]. These tainties were gotten by employing proper Lyapunov
approximations were derived by establishing Jackson functions. Meanwhile, numerical simulations were given
type inequalities, converging the fractional approxima- to demonstrate the effectiveness of the proposed method.
tion results better than the integer-order scheme. Comparison among the proposed controller, classical
The fractional differential polynomial ANN was the backstepping controllers, and radial basis function ANN
proposed method to approximate a multi-parametric terminal sliding mode surface controller was carried
function with polynomials, involving the Caputo deriva- out, showing the proposed controller to have a better
tive. The generalization depended on the Riemann– performance.
Liouville differential operator, and the experimental Finally, fractional power series ANN for solving:
results demonstrated that the approximation to the delay fractional optimal control problems and fractional
exact value with the fractional differential polynomial optimal control problems with equality and inequal-
ANN was quicker than the integer-order method [287].
123
2076
123
Table 3 Control strategies employed to synchronize and stabilize ANN involved with FC
Control strategy References ANN architecture Fractional derivative Application
Adaptive feedback control [224] Competitive FANN Robust synchronization and
stability
[114] Delayed Competitive FANN Global asymptotic stability
[101] Delayed FANN Caputo derivative Projective synchronization
[225] Delayed BAM FANN Global stability
[104] Delayed BAM FANN Synchronization
[184] Delayed memristive FANN Adaptive synchronization
Adaptive feedback control [226] Delayed complex-valued Riemann–Liouville derivative Global synchronization and
FANN global stability
[60] Cohen–Grossberg FANN Exponential stability and
synchronization
[227] Complex-valued FANN Stability and synchronization
[228] Memristive BAM FANN Finite-time impulsive
synchronization
Impulsive Control [229] Memristive discontinuous Caputo derivative Exponential stability
FANN
[65] Quaternion-valued Finite-time stability
memristive FANN
[185] Delayed FANN Global stability and
synchronization
[230] Delayed FANN Almost periodicity and
stability
[186] Delayed cellular FANN Global synchronization
Adaptive impulsive control [231] Quaternion-valued FANN Caputo derivative Stability and impulsive
synchronization
Washout filter control [187] Delayed FANN with Grünwald–Letnikov Chaotic synchronization
derivative
Table 4 ANN involved with FC to reach stability

Application References ANN architecture Fractional derivative
Exponential stability [171] FANN
[32] Complex-valued FANN Caputo derivative
[232] Interval projection FANN
[271] Delayed BAM FANN
Uniform stability [241] FANN
[237] FANN with and without
delays
[76, 242] Delayed FANN
[240] Hopfield FANN
[272] Delayed memristive FANN Caputo derivative
[273] Delayed memristive fuzzy
BAM FANN
[238] Cellular FANN
[79, 233, 234] Delayed complex-valued
FANN
[235, 236] Delayed BAM FANN
Finite-time Stability [248] FANN
[75, 85, 246, 275] Delayed FANN
[247] Delayed FANN
[112, 243, 276] Delayed complex-valued
FANN
memristive FANN
[244] Delayed complex-valued Caputo derivative
memristive FANN
[175] Cohen–Grossberg memristive
FANN
[245] Delayed Cohen–Grossberg
FANN
[89] Delayed Hopfield FANN
Global stability [28] FANN
[59] BAM FANN
[62] Fuzzy FANN
[38, 256] Hopfield FANN
[86] Delayed Hopfield FANN Caputo derivative
[117, 174, 180, 279] Delayed memristive FANN
[106] Delayed Cellular FANN
[185, 258] Delayed cellular FANN
FANN
Global Stability [259] Delayed FANN
[95] Delayed FANN Riemann–Liouville derivative
[280] Delayed fuzzy BAM FANN
[255] Delayed hybrid BAM FANN
Stability and Bifurcation [41] Recurrent FANN
[36, 37] Hopfield FANN
123
Table 4 continued
Application References ANN architecture Fractional derivative
[74, 250, 281] Delayed FANN
[251] Complex-valued Hop-
field FANN
[80] Delayed complex- Caputo derivative
valued Hopfield FANN
[108] Delayed complex-
valued FANN
[77, 249, 282] Delayed BAM FANN
Stability [54, 57, 155, 263, 267] FANN
[268] Delayed FANN
[260] FANN with time-
varying delays
[99] Hopfield ANN
[35, 39, 261] Hopfield FANN
[84, 87] Delayed Hopfield FANN
[41, 66] Recurrent FANN
[262] Nonautonomous FANN Caputo derivative
valued Hopfield FANN
valued memristive
FANN
[94] Delayed quaternion-
valued FANN
Stability [264] Delayed neutral type Riemann–Liouville
FANN derivative
Monostability [283] FANN Caputo derivative
and multista-
bility
Asymptotical [284] Delayed FANN Caputo derivative
stability
Quasi-uniform [252] Delayed FANN Caputo derivative
stability
Fractional [253] FANN Caputo Derivative
input stability
Global stabil- [254] Delayed complex- Caputo derivative
ity and global valued FANN
periodicity
[269] Delayed non-
autonomous FANN
Synchronization [270] Delayed Complex Riemann–Liouville
stability FANN Derivative
Stability and [285] Memristive FANN Caputo derivative
passivity
ity constraints were developed in [47,69] respectively, ANN involved with FC: presence of chaos
involving Caputo derivative, feed-forward ANNs and
according to the Pontryagin minimum principle. In Next, several types of research where ANN involved
both works, the optimization techniques and colloca- with FC have chaotic behaviors will be depicted:
tion methods were proposed to determine the approx- In Ref. [170], a numerical simulation algorithm
imate solution of the fractional optimal control prob- for FDE was presented involving Riemann–Liouville
lems. The obtained results were compared with the derivative, where the chaotic phenomena and their con-
exact solutions and analytical solutions, respectively. trol were discussed by numerical simulation. Chaos
feedback control was developed, allowing to control and
synchronize the FANN system.
123
Moreover, in Ref. [290] obtained the fractional-order Finally, Han in Ref. [200] performed a compos-
model of delayed cellular ANN for describing chaotic ite learning sliding mode control approach to attain
behaviors for fractional-order 0.1 ≤ α < 1 interval. the synchronization of chaotic FANN with unmatched
Meanwhile, delay time values for which the chaos unknown parameters, employing Caputo definition. A
occurred were defined using the largest Lyapunov expo- comparison between the proposed composite learning
nents. Riemann–Liouville derivative was employed in sliding mode control and the common sliding mode con-
this work, and the simulation results demonstrated that trol demonstrated that the proposed composite learning
the time delay where chaos occurred decreased as the sliding mode control establishes an accurate parame-
fractional-order decreased too. ter estimation without the permanent excitation condi-
Moreover, Kaslik and Sivasundaram Seenith in Ref. tion having better control performance than the sliding
[37] investigated the stability, multi-stability bifurca- mode control scheme.
tions, as well as the chaos of Hopfield FANN involving Other research related to chaos of ANN involved with
the Caputo derivative. In this work the critical values of FC applied to image encryption will be found later in
fractional-order where the Hopf bifurcations occurred this paper, [82,186,294] and [295].
were identified, and the stability domain of a steady-
state was characterized. The simulation results demon-
strated that the chaotic behavior appeared when the
fractional-order of the system increased. ANN involved with FC: applications in estimation
In Refs. [31,83], the authors studied the behaviors of
the complex dynamics of a cellular FANN and a delayed The main works about estimation involving ANN and
Hopfield FANN, respectively. To carried out these stud- FC will be depicted in the following summary:
ies, numerical simulations involving the Caputo deriva- On-line state estimation of nonlinear dynamic sys-
tive were developed. The systems investigated in these tems was carried out using Differential FANN. The sim-
works presented dynamic behaviors, such as periodic ulation of two coupled tanks was carried out to demon-
and chaotic motions. Furthermore, in both works, the strate the feasibility of Differential FANN as a nonlinear
existence of chaotic attractors was demonstrated. This systems identifier [296]. In [35], the same author solved
was verified using bifurcation diagrams and phase por- a parameter estimation problem for demonstrating the
traits. Hopfield FAN existence using the Caputo derivative. In
On the other hand, the projective synchronization this case, the stability of Hopfield FANN was reached
of chaotic memristive FANN with time-varying delay applying an energy-like function analysis.
and switching jumps mismatch involving Riemann– Besides, in Ref. [11] authors studied the parameter
Liouville derivative, was studied in [178]; also, in this estimation problem of unknown system parameters on
work, the chaotic behavior of the memristor-based FANN involving Caputo and Riemann–Liouville deriva-
FANN system was showed. tives. Synchronization-based identification method of
Subsequently, Luo et al. in Ref. [156] developed fractional-order was achieved thanks to the combina-
an adaptive synchronization methodology combining tion of adaptive control and parameter update law,
Chebyshev ANN, extended state tracking differentiator, demonstrating the correctness of the obtained results
and adaptive backstepping, to reach the synchroniza- through a numerical example.
tion between the drive system and response system of a Moreover, in Refs. [297,298] were designed a state
fractional-order chaotic arch micro-electro-mechanical estimator and non-fragile state estimator for delayed
system with the uncertain item and time delay under memristive FANN involving the Caputo derivative,
distributed electrostatic actuation. Caputo derivative respectively. In work [297], the state estimators’ exis-
was employed in this work, and the stability of the tence was ensured, and a suitable state estimator for
closed fractional-order arch micro-electromechanical memristive FANN was proposed. Accordingly, based
system was guaranteed based on the fractional-order on the fractional-order Lyapunov direct method, some
Lyapunov stability criterion. The simulation results new sufficient conditions were given to guarantee the
demonstrated the effectiveness of the proposed scheme. existence of the state estimator. On the other hand, in
Moreover, in Ref. [291] stability, bifurcation, and Ref. [298] by using the Lyapunov technique, the authors
chaos of a Memristive FANN with discontinuous mem- were getting sufficient conditions to ensure the global
ductance functions were investigated, employing Caputo asymptotic stability of the error model.
derivative and identifying interesting dynamics, such A radial basis function ANN was used to estimate the
as chaotic motion, tangent bifurcation, and intermit- bound of disturbances in Ref. [299]. And an adaptive
tent chaos. The chaotic attractors were demonstrated fractional sliding mode controller with a neural estima-
to exist over a wide range of some specified parameters. tor for a class of nonlinear systems also was designed.
Otherwise, the chaotic Chua’s attractor of fractional- A robust control law was designed to guarantee
order was studied in Ref. [169]. In Ref. [292] the Lorenz the occurrence of the sliding motion, as well as the
system was studied, the chaotic attractors of Hopfield Hopfield FANN stabilization involving the Riemann–
FANN in [293], memristor-based FANN in Ref. [210], Liouville derivative in Ref. [164]. Moreover, the system’s
two-dimensional delayed FANN in Ref. [101], time- unknown parameters were estimated, and the sliding
delayed inertial FANN in Ref. [220], and nonidentical surface to origin finite-time stability was achieved based
FANN chaotic behaviors were studied in Ref. [163]. on the fractional-order Lyapunov theory. An example of
123
Hopfield FANN was presented to demonstrate the effec- proposed scheme results presented fewer conservation
tiveness of the proposed scheme. effects when comparing with other works found in the
The non-fragile state estimation issue for memristive literature.
BAM FANN with and without time delays was studied
in Ref. [300] by applying the fractional-order derivative ANN involved with FC: applications in periodicity
in Caputo sense. Based on fractional-order Lyapunov
functionals and linear matrix inequalities was ensured The overview of the research works focused on the peri-
the asymptotical stability of the error system. odicity involving FC on ANN will be depicted below:
The quasi-estimation was investigated by Li in Ref. First, in Ref. [71], the authors demonstrated that
[301] employing fuzzy memristive FANN based on the fractional derivative of a periodic function cannot
the Caputo derivative. The quasi-estimation was stud- be a periodic function with the same period involving
ied through a Laplace transformation, and the quasi- Caputo, Riemann–Liouville, and Grünwald–Letnikov
synchronization control was attained due to the designed derivatives; in this paper a FANN was employed to
feedback controller. guarantee the non-existence of periodic solutions in
Besides, in Ref. [104], the authors studied the Mittag– fractional-order dynamical systems.
Leffler state estimator and an adaptive synchronization Subsequently, Wu and Zeng in Ref. [62] derived the
for delayed BAM FANN by applying the Caputo deriva- S-asymptotic ω-periodicity and global asymptotic ω-
tive. An adaptive feedback control was designed, and periodicity of fuzzy FANN, involving Caputo deriva-
Mittag–Leffler adaptive synchronization was reached tive. The difference between integer-order neurody-
using fractional-order inequality techniques. namic systems and fractional-order neurodynamic sys-
Another novel approach in this area is the robust tems was shown. Several simulations were performed to
state estimation of complex-valued FANN with uncer- demonstrate the effectiveness of the proposed method.
tain parameters and BAM FANN with norm-bounded On the other hand, the global stability and global
uncertainties, investigated in Refs. [278,302], respec- asymptotic periodicity for complex-valued FANN with
tively. Hu in Ref. [302], applied the Riemann–Liouville time-varying delays was discussed in Ref. [254], while
derivative, and Nagamani in [278] used the Caputo for non-autonomous FANN with time-varying delays
derivative. Since both systems presented time delays, was addressed in Ref. [269] by applying the Caputo
a new linear matrix inequality criterion was obtained derivative. In these works, the solutions converged to
to reach the asymptotic stability of the systems’ error. the same periodic function. Numerical examples were
In both cases, numerical simulations were performed to given to demonstrate the feasibility of the schemes.
confirm the effectiveness of the proposed schemes. Finally, the global asymptotic ω-periodicity for a
Finally, in Refs. [120,303], the authors investigated non-autonomous FANN involving Caputo derivative
the finite-time projective synchronization of memris- was investigated in [262]. The authors demonstrated
tive FANN with mixed time-varying delays and uncer- that FANN has S-asymptotically periodic solutions.
tain parameters and the finite-time synchronization of Furthermore, all solutions of FANN globally converge
memristive Cohen–Grossberg FANN with time-varying to a periodic function.
delays, respectively. In these works, the Caputo deriva-
tive and feedback controllers were applied. Moreover,
the settling times were estimated. Heat transfer process
Only one work related to the FC on ANN with the heat

ANN involved with FC: applications in dissipativity transfer process will be depicted as follows:
The heat transfer process was modeled using a dis-
Following relevant works related to dissipativity involv- crete FANN involving Grünwald–Letnikov derivative in
ing FC applied to ANN will be shown: Ref. [305]. The experiments and the obtained results
First, in Ref. [100], the authors investigated the showed that the proposed FANN modeled the unknown
global dissipativity of delayed FANN and discontinuous dynamics correctly.
activation functions employing Caputo derivative. In
this research, sufficient conditions were given to ensure ANN involved with FC: applications in sustainable
the dissipativity of the model solution. The effectiveness energy
of the proposed scheme was demonstrated by numerical
examples. ANN involved with FC has been applied in the sustain-
Second, the dissipativity and global asymptotic sta- able energy area, specifically in wind turbine applica-
bility of delayed complex-valued FANN were investi- tions [30,34,149]. These works will be addressed below:
gated in Ref. [257]. In this research, the authors used The behavior of a wind turbine in operation was
the Caputo derivative. Numerical simulations showed identified using a variable order FANN and a Dynamic
the effectiveness of the proposed scheme. FANN involving the Caputo derivative in Refs. [30,34],
Finally, Li in Ref. [304] investigated the dissipa- respectively. The proposed methods were evaluated and
tivity and the exponential synchronization control of validated by using experimental data obtained from the
Memristive FANN involving reaction-diffusion terms, wind turbine under operation. Moreover, in Ref. [34],
the Caputo derivative, and a feedback controller. The results obtained with the dynamic FANN were com-
123
pared with the obtained with the integer-order dynamic filter to establish a robust control for applications
ANN showing that the fractional approach was more of unmanned aerial vehicles (UAVs). Posteriorly, in
accurate. Mobarez et al. the authors implemented a fractional-
Also, in Ref. [149] was presented a fractional-order order PID control based on ANN for fixed-wing UAVs,
PID control implementation for regulating both the where the proposed autopilot was evaluated in lin-
rotor speed and the power production of a wind tur- earized and nonlinear systems. The fractional controller
bine. The control scheme was combined with a radial showed better performance against wind disturbance,
basis function ANN, allowing it to reach better per- the effect of the sensors’ noise, and system uncertain-
formance and robustness than with the integer-order ties when compared with other controllers.
controllers.
ANN involved with FC: applications in medicine

ANN involved with FC: applications in circuits
ANN involved with FC has been applied to the medical
field and reported in Refs. [24,145,197,306]. Below we realization
will describe these research works.
In Ref. [145], the detection of left-sided and right- ANN involved with FC has been applied to circuit
sided hearing loss was carried out using the fractional realization. In works [25,165,189,198,199,292] relevant
Fourier transform and a feedforward ANN trained research have been presented on this issue. Below we
by the Levenberg–Marquardt algorithm. Also, in Ref. will present an overview of these works:
[306] the authors developed a pathological brain detec- In Ref. [165], the authors developed an adaptive frac-
tion system based on multi-layer perceptron ANN tional sliding mode control involving the Caputo deriva-
to improve the interpretation of magnetic resonance tive. The fractional controller was based on the dual
brain images. In this work, the ANN was the classi- radial basis function ANN, allowing to improve the per-
fier that received the fractional Fourier entropy features formance of three-phase shunt active power filters. Also,
extracted from the brain images. The adaptive real- the Chua’s circuit with Caputo derivative was devel-
coded biogeography-based optimization was the algo- oped in Ref. [25] to investigating the fractional Chua’s
rithm implemented to train this ANN, and the pro- system and discuss its chaotic behavior. Furthermore,
posed method was able to interpret the images with in Ref. [189], the authors developed an adaptive ANN
an accuracy of 99.53%, improving the results obtained backstepping control of fractional-order Chua–Hartley
from other pathological brain detection system. chaotic system, as well as an electronic circuit. In this
Subsequently, Sharafian in Ref. [197] used a radial work, the ANN was used to estimate the unknown non-
basis function ANN with a sliding mode observer for linear function, and the proposed controller was able to
modeling the uncertainties of the human immunodefi- guarantee the stability of the closed-loop system. Oth-
ciency virus infection fractional model. The fractional erwise, in Ref. [292], a lathe machine tool was the basis
mathematical model involved the Caputo derivative. to study turning chatter vibration by means of ANN-
The ANN estimated the system uncertainties while the based on Chua’s circuit and fractional-order Lorenz
sliding mode observer eliminated the external distur- master/slave chaotic system. Finally, fractional-order
bances. In this work, the finite-time stability of the sliding mode controllers based on recurrent ANN were
observer was guaranteed. In fact, the radial basis func- developed for the current compensation and the current
tion ANN estimated the complex nonlinearity of the harmonic compensation of active power filter involving
system accurately. Caputo derivative in [198,199], respectively.
Finally, in Ref. [24], the fractional-order Darwinian Also, Liu [308] employed a reaction/diffusion cel-
particle swarm optimization (FO-DPSO) algorithm was lular FANN to describe the diffusion behavior that
employed for training a feed-forward ANN to approxi- happened on the electromagnetic field where the elec-
mate the solutions of the corneal shape model for eye trons describe nonuniform movement. In this work,
surgery. In this work, the Adams–Bashforth–Moulton the Caputo derivative was used, and the stability of
numerical method was used to show the effectiveness of the FANN has achieved thanks to an observed-based
the proposed scheme boundary control.
On the other hand, Ding et al. [309] accomplished
the first sampling-controlled memristive FANN with
ANN involved with FC: applications in unmanned stochastic sensor faults via an impulsive method based
aerial vehicles (UAVs) on Caputo derivative. This novel approach was applied
to the fractional-order Chua’s circuit system, where
ANN involved with FC has been applied to unmanned they reached its stabilization.
aerial vehicles (UAVs). Some relevant works have been Finally, Sanchez et al. [310] simulated the incremen-
reported in Refs. [144,307]. Following, we will address tal capacity curve of an LFP battery model by train-
these works. ing a recurrent FANN. In this work, the model-based
In Ref. [144], the authors approximated the realiza- health prognosis of the LFP battery with an accuracy
tion of an analogical fractional-order PID controller that is comparable with the laboratory measurements
using feed-forward ANN and a finite impulse response was attained.
123
ANN involved with FC: applications in robotic ness of the encryption system. The simulation results
demonstrated that the algorithms have good encryption
ANN involved with FC and Models to control Robots features.
have been developed in Refs. [52,68], and they will be Similar to the research works presented above, in Ref.
summarized as follows: [295], another chaotic image encryption algorithm was
In Ref. [52], the authors proposed a deep convo- proposed. In this case, a five-dimensional cellular FANN
lutional ANN based on the fractional-order sliding was employed as a diffusion controller in the encryp-
mode control scheme to control trajectory tracking of tion system. The results showed that this new algorithm
robotic manipulators involving Caputo derivative. The improves encryption efficiency with good security per-
proposed control showed to have robust performance formance. Furthermore, it resisted the common attack
against parametric uncertainties and external distur- methods.
bances. In this work, several simulations were carried Finally, the synchronization of a class of FANNs was
out to validate the proposed methods. carried out by designing an adaptive control to develop
In Ref. [68] was presented a nonlinear neural a crypto-system algorithm for encryption/decryption of
fractional-order PID controller based on ANN applied unmanned aerial vehicle color images in secure commu-
to the motion control of nonholonomic differential drive nications, employing Caputo derivative [311].
mobile robot, involving Grünwald–Letnikov derivative.
The hybridization of a modified adaptive PSO and ANN involved with FC: applications in other
the EFF optimization algorithms were implemented for engineering applications
tuning the parameters of the fractional-order PID con-
troller based on ANN. The authors showed with numer- Blasik et al. [312] developed an accurate numerical
ical simulations that the performance of the designed method based on ANN and Caputo derivative to extend
fractional controller was excellent compared with non- the front fixing method developed in a previous work
linear controllers of integer-order on the trajectory based on the one phase fractional-order Stefan prob-
tracking of the differential drive mobile robot with dif- lem (anomalous molecular diffusion where the diffusion
ferent trajectories as study cases. coefficient is generalized).
ANN involved with FC: applications in image

encryption 7 Future research topics arising on ANNs
involved with FC
ANN involved with FC have been applied to image
encryption in Refs.: [82,186,294,295]. Such works will The main advantage of modeling systems involving
be overviewed as follows: FANNs is that the fractional derivatives are an excellent
First, in Ref. [186], the authors presented an image tool in describing the memory and hereditary properties
encryption approach based on impulsive synchroniza- of various processes. Therefore, the simulated systems
tion of chaotic ANN applying the fractional-order involving FANNs are more accurate than the integer-
approach involving Caputo derivative and considering order models. However, there are few works done in
a delayed cellular FANN. this field to date, and there are many important top-
Second, image encryption algorithms were designed ics that have not been approached yet. Therefore, we
based on chaotic fuzzy cellular neural FANN with time- would like to recommend some future research topics
varying delays and chaotic three-dimensional discrete where the FANNs could be successfully applied, which
Hopfield FANN in Refs. [82,294], respectively. The are detailed as follows:
authors used the Caputo derivative and the FANNs
as a pseudo-random number generator. The dynamic
• We propose to apply the FANNs to the identifica-
behavior and synchronization of the systems were inves-
tion, control, performance studies, and prediction
tigated and applied to image encryption algorithms
of behaviors on physical models. Thus, the mod-
where solutions allowed to improve encryption security.
els could improve their accuracy without increasing
Third, in [82], numerical evaluations were developed,
their complexity.
and analysis of bifurcation diagrams, phase space dia-
grams, and time series plots to explore the effects of – The employment of FANNS for the simulation of
the time-varying delay and the fractional-order. In this trajectorial physical models could allow that the
work, the global stability conditions were derived under effectiveness of the trajectories prediction to be
the design of an adaptive control approach guarantee- increased, avoiding wasting time and resources
ing the global asymptotic and the exponential stabil- in the experimental studies.
ity by synchronizing the drive-response system when – The management of the plants of industrial pro-
time tends to large. On the other hand, in Ref. [294], cesses is usually complex in nature. Often, the
phase portraits, bifurcation diagrams, and Lyapunov training ANN models must attain a fast response
exponents were developed to show the chaotic dynam- concerning corresponding physical models and
ics of the system. A control approach was designed to for the real-time monitoring of the plants; We
synchronize the system. The results show the effective- propose to use FANNs to model the indus-
123
trial plants since FANNs could do the manage- system as simulated employing FC. Guarantee-
ment of them more efficiently with the fractional ing a good prediction, the forecast of human
approach on their simulations. and economic losses in front of natural disasters
could be employed successfully.
• We propose to apply the FANNs on the sustain-
able energy technologies to realize a better analysis • We propose combining fractional-order physically-
and optimization of their parameter, and thus, help based modeling and deep learning.
environmental care.
– The fractional deep neural networks are a tool of
– FANNs could be an accurate tool for the study machine learning and artificial intelligence more
of behaviors, the prediction of performance, and powerful than integer-order deep neural net-
the optimization of solar collectors, biomass works; the FC improves the deep learning meth-
heating systems, wind turbines, and so on. ods, reaching a better optimization of parame-
– FANNs could be an excellent tool in studying ters of the physical models.
voltage prediction on PEM fuel cells with min-
• We propose the application of FANNs to predict
imum time demand and good accuracy, reduc-
economic trends.
ing costs and avoiding extensive experiments.
FANNs could lead to a better analysis of the – The realization of economic models based on
PEM fuel cell components and better optimize FANNs could help economists and business peo-
their parameters to minimize the voltage losses ple accurately identify entrepreneurship oppor-
and reach their best efficiency. tunities. This approach could help governments
and economists establish correct predictions of
• We propose to employ FANNs on modeling of the
the prices of crude oil, natural gas, power, risk
epidemic spreading of viruses that affect the health
management, trading strategies, etc.
of animals and human beings.
– The employment of FANNs in modeling the
parameters that determine the diagnosis of
COVID-19 would provide fast and accurate diag- 8 Conclusions
nostics of the actual COVID-19 pandemic, help-
ing clinicians in detecting COVID-19, quantifica- This manuscript presented a bibliographic review of
tion, follow-up of the infected cases, and helping fractional calculus (FC) on artificial neural networks
the activation of the plan actions. (ANN). We have focussed on realizing a thorough inves-
– The spreading prediction and analysis of the per- tigation related to the employment of FC on ANN, the
formance of common viruses such as HIV, H1N1, methods used to solve their fractional differential equa-
Dengue, Ebola, and others could be successfully tions (FDE), the optimization algorithms employed to
modeled using FANNs, which is beneficial for train these systems, the control approaches involved
preserving lives since predictions are helpful to with them, and their main applications in different
control and prevent the spread of the viruses. areas of science and engineering. According to the bib-
• We propose to employ FANNs to model the spread liographic review, the most used method to solve the
of cancer cells from where they first formed to FDE on ANNE involved with FC is the analytical
another part of the body. Laplace transform method, followed by the Adams–
Bashforth–Moulton method. Also, other methods have
– The identification of cancer cells and the predic- been used to solve FDE in these systems, such as
tion of their movement into the human body are the Homotopy perturbation method, the power series
essential for preserving life. Thanks to the effec- expansion method, Adomian decomposition method,
tiveness of FANNs on modeling and predicting among others. All of them have allowed establishing
behaviors in different areas of science, we believe a comparison between the proposed schemes and the
that some accurate FANNs models based on can- analytical or numerical solution of FDE to validate
cer cell propagation could help in its identifica- them. Regarding the optimization algorithms used for
tion and prediction. the training of ANN involved with FC, there have
• We propose to apply the FANNs for the simulation been developed a few algorithms with the fractional
of biological processes. approach, indeed, just gradient descent algorithm,
Back-propagation algorithm, and Darwinian particle
– FANNs could be efficiently used in the process swarm optimization algorithm have their fractional ver-
control of biological systems, reaching a better sion. Nevertheless, there are lots of classical algorithms
online optimization of these systems than in the employed with the aim of training ANN involved with
integer-order cases. FC. According to the advantages of FC, most of these
• We propose to simulate probabilistic models using algorithms could be implemented with a fractional
FANNs. approach in the future, obtaining results very differ-
ent from the obtained using classical algorithms and
– The meteorological time series prediction could probably with solutions more near reality. For train-
be more accurate when the ANN describes the ing of these systems, the most used derivative was the
123
Caputo derivative, followed by the Riemann–Liouville good predictions avoiding human and economic losses
derivative, Atangana–Baleanu in Caputo’s sense, and when natural disasters happen. Finally, but not less
Grünwald–Letnikov, respectively. Otherwise, another important, we recommended implementing the FANNs
important topic that has been presented here is the con- based on economic models for guiding governments and
trol of such systems; since several works joined control economists to make decisions based on a correct predic-
strategies with FC to guarantee synchronization and tion of the economic phenomena. All our recommenda-
stabilization of the ANN involved with FC. Concerning tions are based on the certainty that the accurate mod-
the control strategies, the most reported in the litera- els reached employing FANNs avoid the employment
ture is the feedback control, followed by adaptive con- of exhaustive repetitive experimental tests. Thus, the
trol, sliding mode control, adaptive feedback control, FANNs help the growth of science and engineering less
impulsive control, and washout filter control. Therefore, time employing the optimum resources for their devel-
we can affirm that FC is an excellent mathematical tool opment.
to be used with ANN. Also, we have confirmed that the
Caputo derivative is the most applied with ANN due to
Acknowledgements Edumis Viera Martin acknowledge
its ability to describe physical systems. The Riemann–
the support provided by CONACyT through the assignment
Liouville derivative is the second derivative most used doctoral fellowship. Jesús Emmanuel Solı́s Pérez acknowl-
on these types of systems. This derivative is applied edges the support provided by CONACyT through the
to describe theoretical systems, while the Grünwald– assignment post-doctoral fellowship. José Francisco Gómez
Letnikov derivative is less used in these types of sys- Aguilar acknowledges the support provided by CONA-
tems. Other interesting applications of ANN involved CyT: cátedras CONACyT para jóvenes investigadores 2014.
with FC are the approximation of functions, the dissipa- José Francisco Gómez Aguilar, Jesús Emmanuel Solı́s
tivity, periodicity demonstration, description of chaotic Pérez, José Alfredo Hernández-Pérez and Ricardo Fabri-
behaviors, among others. But some of the most impor- cio Escobar-Jiménez acknowledges the support provided by
tant advances related to ANN involved with FC are SNI-CONACyT.
their recent applications in medicine, robotic, cryptog-
raphy, image processing, and sustainable energy. This Declarations
implies that the research community is paying attention
to the fractional calculus theory.
We are in the presence of an interesting branch of Conflict of interest The authors declare no conflict of
mathematics compared with integer-order calculus. The interest.
FC can better describe several processes’ memory and
genetic characteristics, having unlimited memory and
more degrees of freedom.
However, few applications exist until today; we con-
sider many promising works to do shortly. Therefore, References
we decided to guide the researchers for the realization
of future works in this field. Some of the future direc- 1. Z. Wang, X. Wang, Y. Li, X. Huang, Stability and Hopf
tions we suggested are the following: the application bifurcation of fractional-order complex-valued single
of FANNs on the study of trajectorial physical models neuron model with time delay. Int. J. Bifurc. Chaos
for reach more accurate simulated systems and better 27(13), 1750209 (2018)
prediction of trajectories than in integer-order cases; 2. R. Rakkiyappan, G. Velmurugan, F.A. Rihan, S.
the employment of FANNs for the improvement of the Lakshmanan, Stability analysis of memristor-based
voltage prediction in PEM fuel cells and better identifi- complex-valued recurrent neural networks with time
cation of their parameters; the simulation of industrial delays. Complexity 21(4), 14–39 (2016)
plants using FANNs to reach more accurate computa- 3. C.J. Zúñiga-Aguilar, H.M. Romero-Ugalde, J.F.
tional systems which can be managed on real-time with Gómez-Aguilar, R.F. Escobar-Jiménez, M. Valtierra-
fast-response. In medicine and biology, we proposed Rodrı́guez, Solving fractional differential equations of
implementing FANNs based on the epidemic spread- variable-order involving operators with Mittag–Leffler
ing of viruses such as Covid-19, HIV, Ebola, H1N1, kernel using artificial neural networks. Chaos Solitons
and others for a better analysis, diagnosis, prediction, Fract. 103, 382–403 (2017)
and forecast of them. Also, we proposed to realize the 4. S. Qureshi, A. Yusuf, A. Shaikh, M. Inc, Transmission
dynamics of varicella zoster virus modeled by classi-
simulation of cancer cells spreading into the human
cal and novel fractional operators using real statistical
body to help clinicians detect, study the behavior of
data. Phys. A 534, 1–12 (2019)
cancer cells, and predict their performance accurately, 5. A. Jajarmi, A. Yusuf, D. Baleanu, M. Inc, A new frac-
helping save lives. Besides, we proposed implementing tional HRSV model and its optimal control: a non-
FANNs on the biology process to aid the development singular operator approach. Phys. A 547, 1–12 (2020)
of process control of biological systems and predict the 6. S. Qureshi, A. Yusuf, A. Ali Shaikh, M. Inc, D.
performance of these systems, avoiding realize exhaus- Baleanu, Mathematical modeling for adsorption pro-
tive failed experiments. On the other hand, we recom- cess of dye removal nonlinear equation using power law
mended applying FANNs on probabilistic meteorologi- and exponentially decaying kernels. Chaos Interdiscip.
cal models to attain accurate systems capable of making J. Nonlinear Sci. 30, 1–10 (2020)
123
7. A. Khan, G. Hussain, M. Inc, G. Zaman, Existence, tion. New Astron. 75(2019), 101307 (2020). https://
uniqueness, and stability of fractional hepatitis B epi- doi.org/10.1016/j.newast.2019.101307
demic model. Chaos Interdiscip. J. Nonlinear Sci. 30, 22. A.H. HadianRasanan, N. Bajalan, K. Parand, J.A.
1–10 (2020) Rad, Simulation of nonlinear fractional dynamics aris-
8. M. Inc, N. Bouteraa, M. Akinlar, S. Benaicha, Y. Chu, ing in the modeling of cognitive decision making using
G. Weber, B. Almohsen, New positive solutions of non- a new fractional neural network. Math. Methods Appl.
linear elliptic PDEs. Appl. Sci. 10, 1–13 (2020) Sci. 43(3), 1437–1466 (2020)
9. A. Yusuf, B. Acay, U. Mustapha, M. Inc, D. Baleanu, 23. M. Syed Ali, G. Narayanan, V. Shekher, A. Alsaedi,
Mathematical modeling of pine wilt disease with B. Ahmad, Global Mittag–Leffler stability analysis of
Caputo fractional operator. Chaos Solitons Fract. 143, impulsive fractional-order complex-valued BAM neural
1–13 (2021) networks with time varying delays. Commun. Nonlin-
10. B. Acay, M. Inc, Fractional modeling of temperature ear Sci. Numer. Simul. 83, 105088 (2020). https://doi.
dynamics of a building with singular kernels. Chaos org/10.1016/j.cnsns.2019.105088
Solitons Fract. 142, 1–11 (2021) 24. W. Waseem, M. Sulaiman, A. Alhindi, H. Alhakami,
11. Y. Gu, Y. Yu, H. Wang, Synchronization-based param- A soft computing approach based on fractional order
eter estimation of fractional-order neural networks. DPSO algorithm designed to solve the corneal model
Phys. A 483, 351–361 (2017). https://doi.org/10.1016/ for eye surgery. IEEE Access 8(c), 61576–61592 (2020)
j.physa.2017.04.124 25. N.A. Khan, T. Hameed, O.A. Razzaq, M. Ayaz, Track-
12. M. Asif, Z. Raja, R. Samar, M. Anwar, Design of unsu- ing the chaotic behaviour of fractional-order Chua’s
pervised fractional neural network model optimized system by Mexican hat wavelet-based artificial neural
with interior point algorithm for solving Bagley–Torvik network. J. Low Freq. Noise Vib. Act. Control 38(3–4),
equation. Math. Comput. Simul. 132, 139–158 (2017). 1279–1296 (2018)
https://doi.org/10.1016/j.matcom.2016.08.002 26. M. Pakdaman, A. Ahmadian, S. Effati, S. Salahshour,
13. J. Yu, C. Hu, H. Jiang, X. Fan, Projective synchroniza- D. Baleanu, Solving differential equations of frac-
tion for fractional neural networks. Neural Netw. 49, tional order using an optimization technique based on
87–95 (2014). https://doi.org/10.1016/j.neunet.2013. training artificial neural network. Appl. Math. Com-
10.002 put. 293, 81–95 (2017). https://doi.org/10.1016/j.amc.
14. L.M. Wang, Q.K. Song, Z.J. Zhao, Global asymptotic 2016.07.021
stability of memristor-based fractional-order complex- 27. Jing, Y., Dong, H., Liang, G.: Study on charac-
valued neural networks with time delays. Appl. Math. teristic of fractional master-slave neural network, in
Mech. 38(3), 333–346 (2017). https://doi.org/10.1016/ Proceedings—2012 5th International Symposium on
j.neucom.2017.02.086 Computational Intelligence and Design, ISCID 2012,
15. Z. Ding, Y. Shen, Projective synchronization of non- vol. 2. (IEEE, 2012), pp. 498–501
identical fractional-order neural networks based on 28. S. Zhang, Y. Yu, J. Yu, LMI conditions for global sta-
sliding mode controller. Neural Netw. 76, 97–105 bility of fractional-order neural networks. IEEE Trans.
(2016). https://doi.org/10.1016/j.neunet.2016.01.006 Neural Netw. Learn. Syst. 28(10), 2423–2433 (2017)
16. A. Wu, Z. Zeng, Global Mittag–Leffler stabilization 29. Z. Yang, J. Li, X. Tang, Y. Niu, Hybrid projective syn-
of fractional-order memristive neural networks. IEEE chronization of fractional-order neural networks with
Trans. Neural Netw. Learn. Syst. 28(1), 206–217 different dimensions. J. Phys. Soc. Jpn. 88(8), 1–6
(2017) (2019)
17. I. Podlubny, in Fractional Differential Equations: An 30. Z. Aslipour, A. Yazdizadeh, Identification of nonlinear
Introduction to Fractional Derivatives, Fractional Dif- systems using adaptive variable-order fractional neu-
ferential Equations, to Methods of Their Solution and ral networks (case study: a wind turbine with practi-
Some of Their Applications, vol. 198 (Elsevier, 1998) cal results). Eng. Appl. Artif. Intell. 85(2018), 462–
18. A. Atangana, D. Baleanu, New fractional derivatives 473 (2019). https://doi.org/10.1016/j.engappai.2019.
with nonlocal and non-singular kernel: theory and 06.025
application to heat transfer model. Therm. Sci. 20, 31. X. Huang, Z. Zhao, Z. Wang, Y. Li, Chaos and
763–769 (2016) hyperchaos in fractional-order cellular neural networks.
19. K. Rajagopal, M. Tuna, A. Karthikeyan, s Koyuncu, Neurocomputing 94, 13–21 (2012). https://doi.org/10.
P. Duraisamy, A. Akgul, Dynamical analysis, sliding 1016/j.neucom.2012.01.011
mode synchronization of a fractional-order memristor 32. J. Jian, P. Wan, Lagrange α-exponential stability
Hopfield neural network with parameter uncertainties and α-exponential convergence for fractional-order
and its non-fractional-order FPGA implementation. complex-valued neural networks. Neural Netw. 91,
Eur. Phys. J. Spec. Top. 228(10), 2065–2080 (2019) 1–10 (2017). https://doi.org/10.1016/j.neunet.2017.03.
20. X. Yang, C. Li, Q. Song, J. Chen, J. Huang, Global 011
Mittag–Leffler stability and synchronization analysis 33. Z. Aslipour, A. Yazdizadeh, Identification of Dama-
of fractional-order quaternion-valued neural networks vand tokamak using fractional order dynamic neural
with linear threshold neurons. Neural Netw. 105, 88– network. Trans. Inst. Meas. Control. 41(5), 1447–1457
103 (2018). https://doi.org/10.1016/j.neunet.2018.04. (2019)
015 34. Aslipour, Yazdizadeh, Identification of wind turbine
21. A.H. Hadian-Rasanan, D. Rahmati, S. Gorgin, K. using fractional order dynamic neural network and
Parand, A single layer fractional orthogonal neural net- optimization algorithm. Int. J. Eng. 33(2), 277–284
work for solving various types of Lane–Emden equa- (2020)
123
35. A. Boroomand, M.B. Menhaj, Fractional-order Hop- 49. Yavari, Nazemi, Fractional infinite-horizon optimal
field neural networks, in Lecture Notes in Computer control problems with a feed forward neural network
Science (including subseries Lecture Notes in Artificial scheme. Netw. Comput. Neural Syst. 30(1–4), 125–
Intelligence and Lecture Notes in Bioinformatics), vol. 147 (2019). https://doi.org/10.1080/0954898X.2019.
5506 LNCS, no. PART 1 (2009), pp. 883–890 1688878
36. E. Kaslik, S. Sivasundaram, Dynamics of fractional- 50. H. Antil, R. Khatri, R.L. Lohner, D. Verma, Frac-
order neural networks, in The 2011 International Joint tional deep neural network via constrained optimiza-
Conference on Neural Networks. (IEEE, 2011), pp. tion. Mach. Learn. Sci. Technol. 2, 1–21 (2020)
611–618 51. N. Sadati, A. Ghaffarkhah, S. Ostadabbas, A new neu-
37. E. Kaslik, S. Sivasundaram Seenith, Nonlinear dynam- ral network based FOPID controller, in Proceedings of
ics and chaos in fractional-order neural networks. Neu- 2008 IEEE International Conference on Networking,
ral Netw. 32, 245–256 (2012). https://doi.org/10.1016/ Sensing and Control, ICNSC (2008), pp. 762–767
j.neunet.2012.02.030 52. M. Zhou, Y. Feng, C. Xue, F. Han, Deep convolutional
38. S. Zhang, Y. Yu, W. Hu, Robust stability analysis of neural network based fractional-order terminal sliding-
fractional-order Hopfield neural networks with parame- mode control for robotic manipulators. Neurocomput-
ter uncertainties. Math. Probl. Eng. 2014, 1–22 (2014) ing (2019). https://doi.org/10.1016/j.neucom.2019.04.
39. C.A. Tavares, T.M. Santos, N.H. Lemes, J.P. dos San- 087 (online)
tos, J.C. Ferreira, J.P. Braga, Solving ill-posed prob- 53. L. Chen, J. Qu, Y. Chai, R. Wu, G. Qi, Synchronization
lems faster using fractional-order Hopfield neural net- of a class of fractional-order chaotic neural networks.
work. J. Comput. Appl. Math. 381, 112984 (2020). Entropy 15(8), 3265–3276 (2013)
https://doi.org/10.1016/j.cam.2020.112984 54. M. Gai, S. Cui, S. Liang, X. Liu, Frequency distributed
40. J. Jia, Z. Zeng, LMI-based criterion for global Mittag– model of Caputo derivatives and robust stability of
Leffler lag quasi-synchronization of fractional-order a class of multi-variable fractional-order neural net-
memristor-based neural networks via linear feedback works with uncertainties. Neurocomputing 202, 91–97
pinning control. Neurocomputing 412, 226–243 (2020). (2016). https://doi.org/10.1016/j.neucom.2016.03.043
https://doi.org/10.1016/j.neucom.2020.05.074 55. H. Liu, S.-G. Li, H.-X. Wang, G.-J. Li, Adaptive
41. M. Xiao, W.X. Zheng, G. Jiang, J. Cao, Undamped fuzzy synchronization for a class of fractional-
oscillations generated by Hopf bifurcations in order neural networks. Chin. Phys. B 26(3),
fractional-order recurrent neural networks with 030504 (2017). [Online]. http://www.stacks.iop.
Caputo derivative. IEEE Trans. Neural Netw. Learn. org/1674-1056/26/i=3/a=030504?key=crossref.
Syst. 26(12), 3201–3214 (2015) de3f8b49032fccf86e10604042b47319
42. M.A.Z. Raja, I.M. Qureshi, J.A. Khan, Swarm intelli- 56. G. Li, H. Liu, Stability analysis and synchronization
gence optimized neural networks for solving fractional for a class of fractional-order neural networks. Entropy
differential equations. Int. J. Innov. Comput. Inf. Con- 18(2), 1–13 (2016)
trol 7(11), 6301–6318 (2011) 57. X. Yang, C. Li, Q. Song, T. Huang, X. Chen, Mittag–
43. T. Hu, X. Zhang, S. Zhong, Global asymptotic synchro- Leffler stability analysis on variable-time impul-
nization of nonidentical fractional-order neural net- sive fractional-order neural networks. Neurocomput-
works. Neurocomputing 313, 39–46 (2018). https:// ing 207, 276–286 (2016). https://doi.org/10.1016/j.
doi.org/10.1016/j.neucom.2018.05.098 neucom.2016.04.045
44. M.A.Z. Raja, M.A. Manzar, R. Samar, An efficient 58. H. Liu, Y. Pan, S. Li, Y. Chen, Synchronization
computational intelligence approach for solving frac- for fractional-order neural networks with full/under-
tional order Riccati equations using ANN and SQP. actuation using fractional-order sliding mode control.
Appl. Math. Model. 39(10–11), 3075–3093 (2015). Int. J. Mach. Learn. Cybern. 9(7), 1219–1232 (2018)
https://doi.org/10.1016/j.apm.2014.11.024 59. A. Wu, Z. Zeng, X. Song, Global Mittag–Leffler sta-
45. H. Jahanbakhti, A novel fractional-order neural net- bilization of fractional-order bidirectional associative
work for model reduction of large-scale systems with memory neural networks. Neurocomputing 177, 489–
fractional-order nonlinear structure. Soft. Comput. 496 (2016)
24(17), 13 489-13 499 (2020). https://doi.org/10.1007/ 60. S. Yang, C. Hu, J. Yu, H. Jiang, Exponential stabil-
s00500-020-04763-5 ity of fractional-order impulsive control systems with
46. M.A.Z. Raja, J.A. Khan, I.M. Qureshi, A new stochas- applications in synchronization. IEEE Trans. Cybern.
tic approach for solution of Riccati differential equa- 50(7), 3157–3168 (2020)
tion of fractional order. Ann. Math. Artif. Intell. 60(3), 61. S. Yang, J. Yu, C. Hu, H. Jiang, Quasi-projective syn-
229–250 (2010) chronization of fractional-order complex-valued recur-
47. S. Ghasemi, A. Nazemi, A fractional power series neu- rent neural networks. Neural Netw. 104, 104–113
ral network for solving a class of fractional optimal con- (2018). https://doi.org/10.1016/j.neunet.2018.04.007
trol problems with equality and inequality constraints. 62. A. Wu, Z. Zeng, Boundedness, Mittag–Leffler stabil-
Netw. Comput. Neural Syst. 30(1–4), 148–175 (2019). ity and asymptotical ω-periodicity of fractional-order
https://doi.org/10.1080/0954898X.2019.1693647 fuzzy neural networks. Neural Netw. 74, 73–84 (2016).
48. M. Yavari, A. Nazemi, An efficient numerical scheme https://doi.org/10.1016/j.neunet.2015.11.003
for solving fractional infinite-horizon optimal control 63. S. Yang, J. Yu, C. Hu, H. Jiang, Finite-time synchro-
problems. ISA Trans. 94, 108–118 (2019). https://doi. nization of memristive neural networks with fractional-
org/10.1016/j.isatra.2019.04.016 order. IEEE Trans. Syst. Man Cybern. Syst. PP, 1–12
(2019)
123
64. D. Lin, X. Chen, B. Li, X. Yang, LMI condi- ral Process. Lett. (2019). https://doi.org/10.1007/
tions for some dynamical behaviors of fractional- s11063-018-9801-0
order quaternion-valued neural networks. Adv. Dif- 79. R. Rakkiyappan, J. Cao, G. Velmurugan, Existence
fer. Equ. 1, 2019 (2019). https://doi.org/10.1186/ and uniform stability analysis of fractional-order
s13662-019-2207-0 complex-valued neural networks with time delays.
65. A. Pratap, R. Raja, J. Alzabut, J. Dianavinnarasi, IEEE Trans. Neural Netw. Learn. Syst. 26(1), 84–97
J. Cao, G. Rajchakit, Finite-time Mittag–Leffler sta- (2015)
bility of fractional-order quaternion-valued memris- 80. R. Rakkiyappan, K. Udhayakumar, G. Velmurugan, J.
tive neural networks with impulses. Neural Process. Cao, A. Alsaedi, Stability and Hopf bifurcation analy-
Lett. 51(2), 1485–1526 (2020). https://doi.org/10. sis of fractional-order complex-valued neural networks
1007/s11063-019-10154-1 with time delays. Adv. Differ. Equ. 1, 2017 (2017).
66. P. Liu, Z. Zeng, J. Wang, Multiple Mittag–Leffler sta- https://doi.org/10.1186/s13662-017-1266-3
bility of fractional-order recurrent neural networks. 81. J. Yuan, L. Zhao, C. Huang, M. Xiao, Novel results on
IEEE Trans. Syst. Man Cybern. Syst. 47(8), 2279–2288 bifurcation for a fractional-order complex-valued neu-
(2017) ral network with leakage delay. Phys. A 514, 868–883
67. W. Li, Design and implement of neural network based (2019). https://doi.org/10.1016/j.physa.2018.09.138
fractional-order controller. in Robotic Welding, Intel- 82. P. Mani, R. Rajan, L. Shanmugam, Y. Hoon Joo,
ligence and Automation (Springer, Berlin, Heidelberg, Adaptive control for fractional order induced chaotic
2007), pp. 471–479 fuzzy cellular neural networks and its application to
68. G.A.R. Ibraheem, A.T. Azar, I.K. Ibraheem, A.J. image encryption. Inf. Sci. 491, 74–89 (2019)
Humaidi, A novel design of a neural network-based 83. X. Huang, Z. Wang, Y. Li, Nonlinear dynamics and
fractional PID controller for mobile robots using chaos in fractional-order Hopfield neural networks with
hybridized fruit fly and particle swarm optimization. delay. Adv. Math. Phys. 22, 1–16 (2013)
Complexity 1, 1–14 (2020) 84. H. Wang, Y. Yu, G. Wen, Stability analysis of
69. F. Kheyrinataj, A. Nazemi, Fractional power series fractional-order Hopfield neural networks with time
neural network for solving delay fractional optimal delays. Neural Netw. 55, 98–109 (2014). https://doi.
control problems. Connect. Sci. 32(1), 53–80 (2020). org/10.1016/j.neunet.2014.03.012
https://doi.org/10.1080/09540091.2019.1605498 85. R. Wu, Y. Lu, L. Chen, Finite-time stability of
70. M. Ahmad, E. Al-Solami, Evolving dynamic s-boxes fractional delayed neural networks. Neurocomputing
using fractional-order Hopfield neural network based 149(PB), 700–707 (2015). https://doi.org/10.1016/j.
scheme. Entropy 22(7), 717 (2020) neucom.2014.07.060
71. E. Kaslik, S. Sivasundaram, Non-existence of periodic 86. H. Wang, Y. Yu, G. Wen, S. Zhang, J. Yu, Global sta-
solutions in fractional-order dynamical systems and a bility analysis of fractional-order Hopfield neural net-
remarkable difference between integer and fractional- works with time delay. Neurocomputing 154, 15–23
order derivatives of periodic functions. Nonlinear Anal. (2015). https://doi.org/10.1016/j.neucom.2014.12.031
Real World Appl. 13(3), 1489–1497 (2012). https:// 87. H. Wang, Y. Yu, G. Wen, S. Zhang, Stability analy-
doi.org/10.1016/j.nonrwa.2011.11.013 sis of fractional-order neural networks with time delay.
72. I. Petráš, A note on the fractional-order cellular neural Neural Process. Lett. 42(2), 479–500 (2015)
networks, in IEEE International Conference on Neural 88. H.P. Hu, J.K. Wang, F.L. Xie, Dynamics analysis of
Networks—Conference Proceedings (IEEE), pp. 1021– a new fractional-order Hopfield neural network with
1024 (2006) delay and its generalized projective synchronization.
73. V. Çelik, Bifurcation analysis of fractional order single Entropy 21(1), 1–12 (2019)
cell with delay. Int. J. Bifurc. Chaos 25(2), 1–11 (2015) 89. F. Du, J.-G. Lu, New criteria on finite-time stability
74. C. Huang, J. Cao, Z. Ma, Delay-induced bifurcation in of fractional-order Hopfield neural networks with time
a tri-neuron fractional neural network. Int. J. Syst. Sci. delays. IEEE Trans. Neural Netw. Learn. Syst. 1, 1–9
47(15), 3668–3677 (2016) (2020)
75. X. Yang, Q. Song, Y. Liu, Z. Zhao, Finite-time stability 90. G. Velmurugan, R. Rakkiyappan, Hybrid projective
analysis of fractional-order neural networks with delay. synchronization of fractional-order neural networks
Neurocomputing 152, 19–26 (2015). https://doi.org/ with time delays, in Springer Proceedings in Mathemat-
10.1016/j.neucom.2014.11.023 ics and Statistics, vol. 143(1) (Springer Netherlands,
76. S.M. Abedi Pahnehkolaei, A. Alfi, J.A. Machado, Uni- 2015), pp. 645–655
form stability of fractional order leaky integrator echo 91. X. Huang, Y. Fan, J. Jia, Z. Wang, Y. Li, Quasi-
state neural network with multiple time delays. Inf. Sci. synchronisation of fractional-order memristor-based
418–419, 703–716 (2017). https://doi.org/10.1016/j. neural networks with parameter mismatches. IET Con-
ins.2017.08.046 trol Theory Appl. 11(14), 2317–2327 (2017)
77. C. Huang, J. Cao, Impact of leakage delay on bifur- 92. X. Yang, C. Li, T. Huang, Q. Song, X. Chen, Quasi-
cation in high-order fractional BAM neural networks. uniform synchronization of fractional-order memristor-
Neural Netw. 98, 223–235 (2018). https://doi.org/10. based neural networks with delay. Neurocomput-
1016/j.neunet.2017.11.020 ing 234, 205–215 (2017). https://doi.org/10.1016/j.
78. R. Ye, X. Liu, H. Zhang, J. Cao, Global Mittag– neucom.2017.01.014
Leffler synchronization for fractional-order BAM neu- 93. J. Xiao, S. Zhong, Synchronization and stability of
ral networks with impulses and multiple variable delayed fractional-order memristive quaternion-valued
delays via delayed-feedback control strategy. Neu-
123
neural networks with parameter uncertainties. Neuro- 107. A. Pratap, R. Raja, J. Cao, C.P. Lim, O. Bagdasar,
computing 363, 321–338 (2019) Stability and pinning synchronization analysis of frac-
94. S.M.A. Pahnehkolaei, A. Alfi, J.A. Machado, Delay tional order delayed Cohen–Grossberg neural networks
independent robust stability analysis of delayed frac- with discontinuous activations. Appl. Math. Comput.
tional quaternion-valued leaky integrator echo state 359, 241–260 (2019). https://doi.org/10.1016/j.amc.
neural networks with QUAD condition. Appl. Math. 2019.04.062
Comput. 359, 278–293 (2019). https://doi.org/10. 108. C. Huang, J. Cao, M. Xiao, A. Alsaedi, T. Hayat, Bifur-
1016/j.amc.2019.04.083 cations in a delayed fractional complex-valued neural
95. H. Zhang, R. Ye, J. Cao, A. Ahmed, X. Li, Y. Wan, network. Appl. Math. Comput. 292, 210–227 (2017).
Lyapunov functional approach to stability analysis https://doi.org/10.1016/j.amc.2016.07.029
of Riemann–Liouville fractional neural networks with 109. H.L. Li, C. Hu, J. Cao, H. Jiang, A. Alsaedi, Quasi-
time-varying delays. Asian J. Control 20(5), 1938–1951 projective and complete synchronization of fractional-
(2018) order complex-valued neural networks with time
96. H. Zhang, M. Ye, J. Cao, A. Alsaedi, Synchroniza- delays. Neural Netw. 118, 102–109 (2019). https://doi.
tion Control of Riemann–Liouville fractional competi- org/10.1016/j.neunet.2019.06.008
tive network systems with time-varying delay and dif- 110. X. Wang, Z. Wang, X. Zhu, B. Meng, J. Xia, Stabil-
ferent time scales. Int. J. Control Autom. Syst. 16(3), ity and Hopf bifurcation of fractional-order complex-
1404–1414 (2018) valued neural networks with time-delay. IEEE Access
97. C.J. Zúñiga-Aguilar, A. Coronel-Escamilla, J.F. 7, 158 798-158 807 (2019)
Gómez-Aguilar, V.M. Alvarado-Martı́nez, H.M. 111. X. You, Q. Song, Z. Zhao, Global Mittag–Leffler sta-
Romero-Ugalde, New numerical approximation for bility and synchronization of discrete-time fractional-
solving fractional delay differential equations of order complex-valued neural networks with time delay.
variable order using artificial neural networks. Eur. Neural Netw. 122, 382–394 (2020). https://doi.org/10.
Phys. J. Plus 133(2), 1–6 (2018) 1016/j.neunet.2019.11.004
98. F. Kheyrinataj, A. Nazemi, Fractional Chebyshev func- 112. T. Hu, Z. He, X. Zhang, S. Zhong, Finite-time stabil-
tional link neural network-optimization method for ity for fractional-order complex-valued neural networks
solving delay fractional optimal control problems with with time delay. Appl. Math. Comput. 365, 124715
Atangana–Baleanu derivative. Optim. Control Appl. (2020). https://doi.org/10.1016/j.amc.2019.124715
Methods 41(3), 808–832 (2020) 113. R. Rakkiyappan, G. Velmurugan, X. Li, Complete sta-
99. N.E. Tatar, Fractional Halanay inequality of order bility analysis of complex-valued neural networks with
between one and two and application to neural network time delays and impulses. Neural Process. Lett. 41(3),
systems. Adv. Differ. Equ. 1, 2019 (2019). https://doi. 435–468 (2015). https://doi.org/10.1016/j.chaos.2015.
org/10.1186/s13662-019-2208-z 08.003
100. Z. Ding, Y. Shen, Global dissipativity of fractional- 114. A. Pratap, R. Raja, J. Cao, G. Rajchakit, H.M. Far-
order neural networks with time delays and discontinu- doun, Stability and synchronization criteria for frac-
ous activations. Neurocomputing 196, 159–166 (2016). tional order competitive neural networks with time
https://doi.org/10.1016/j.neucom.2016.03.005 delays: an asymptotic expansion of Mittag Leffler
101. W. Zhang, J. Cao, R. Wu, A. Alsaedi, F.E. Alsaadi, function. J. Frankl. Inst. 356(4), 2212–2239 (2019).
Projective synchronization of fractional-order delayed https://doi.org/10.1016/j.jfranklin.2019.01.017
neural networks based on the comparison principle. 115. A. Pratap, R. Raja, R.P. Agarwal, J. Cao, O. Bag-
Adv. Differ. Equ. 2018(1), 1–16 (2018). https://doi. dasar, Multi-weighted complex structure on fractional
org/10.1186/s13662-018-1530-1 order coupled neural networks with linear coupling
102. G.C. Wu, T. Abdeljawad, J. Liu, D. Baleanu, K.T. Wu, delay: a robust synchronization problem. Neural Pro-
Mittag–Leffler stability analysis of fractional discrete- cess. Lett. 51(3), 2453–2479 (2020). https://doi.org/
time neural networks via fixed point technique. Non- 10.1007/s11063-019-10188-5
linear Anal. Model. Control 24(6), 919–936 (2019) 116. G. Velmurugan, R. Rakkiyappan, J. Cao, Finite-time
103. Y. Cao, C. Bai, Finite-time stability of fractional-order synchronization of fractional-order memristor-based
bam neural networks with distributed delay. Abstr. neural networks with time delays. Neural Netw. 73,
Appl. Anal. 2014, 1–16 (2014) 36–46 (2016). https://doi.org/10.1016/j.neunet.2015.
104. A. Pratap, R. Raja, G. Rajchakit, J. Cao, O. Bagdasar, 09.012
Mittag–Leffler state estimator design and synchroniza- 117. L. Chen, J. Cao, R. Wu, J.A. Tenreiro Machado,
tion analysis for fractional-order BAM neural networks A.M. Lopes, H. Yang, Stability and synchronization of
with time delays. Int. J. Adapt. Control Signal Process. fractional-order memristive neural networks with mul-
33(5), 855–874 (2019) tiple delays. Neural Netw. 94, 76–85 (2017). https://
105. Z. Yang, J. Zhang, Stability analysis of fractional-order doi.org/10.1016/j.neunet.2017.06.012
bidirectional associative memory neural networks with 118. L. Zhang, Y. Yang, F. Wang, X. Sui, Lag synchro-
mixed time-varying delays. Complexity 1, 1–17 (2019) nization for fractional-order memristive neural net-
106. H. Qu, T. Zhang, J. Zhou, Global stability analysis works with time delay via switching jumps mismatch.
of S-asymptotically ω-periodic oscillation in fractional- J. Frankl. Inst. 355(3), 1217–1240 (2018). https://doi.
order cellular neural networks with time variable org/10.1016/j.jfranklin.2017.12.017
delays. Neurocomputing 399, 390–398 (2020). https:// 119. J. Xiao, S. Zhong, Y. Li, F. Xu, Finite-time Mittag–
doi.org/10.1016/j.neucom.2020.03.005 Leffler synchronization of fractional-order memristive
BAM neural networks with time delays. Neurocom-
123
puting 219(2016), 431–439 (2017). https://doi.org/10. 134. D. Sheng, Y. Wei, Y. Chen, Y. Wang, Convolutional
1016/j.neucom.2016.09.049 neural networks with fractional order gradient method.
120. M. Hui, C. Wei, J. Zhang, H.H.C. Iu, N. Luo, R. Neurocomputing 408, 42–50 (2019)
Yao, L. Bai, Finite-time synchronization of memristor- 135. M.R. Chen, B.P. Chen, G.Q. Zeng, K.D. Lu, P. Chu,
based fractional order Cohen–Grossberg neural net- An adaptive fractional-order BP neural network based
works. IEEE Access 8, 73 698-73 713 (2020) on extremal optimization for handwritten digits recog-
121. C. Huang, X. Nie, X. Zhao, Q. Song, Z. Tu, M. Xiao, J. nition. Neurocomputing 391, 260–272 (2020)
Cao, Novel bifurcation results for a delayed fractional- 136. C.J. Zuñiga Aguilar, J.F. Gómez-Aguilar, V.M.
order quaternion-valued neural network. Neural Netw. Alvarado-Martı́nez, H.M. Romero-Ugalde, Fractional
117, 67–93 (2019). https://doi.org/10.1016/j.neunet. order neural networks for system identification. Chaos
2019.05.002 Solitons Fract. 130, 109444 (2020)
122. S. Thanompolkrang, W. Sawangtong, P. Sawangtong, 137. Y.F. Pu, Z. Yi, J.L. Zhou, Fractional Hopfield neural
Application of the generalized Laplace homotopy per- networks: fractional dynamic associative recurrent neu-
turbation method to the time-fractional black-scholes ral networks. IEEE Trans. Neural Netw. Learn. Syst.
equations based on the Katugampola fractional deriva- 28(10), 2319–2333 (2017)
tive in Caputo type. Computation 9(3), 33 (2021) 138. Y.-F. PU, J. Wang, Fractional-order backpropaga-
123. A. Alomari, G.A. Drabseh, M.F. Al-Jamal, R.B. tion neural networks: modified fractional-order steepest
AlBadarneh, Numerical simulation for fractional phi- descent method for family of backpropagation neural
4 equation using homotopy Sumudu approach. Int. J. networks, 1–18 (2019). arXiv:1906.09524
Simul. Process Model. 16(1), 26–33 (2021) 139. A. Sharafian, R. Ghasemi, Fractional neural observer
124. M. Johansyah, A. Supriatna, E. Rusyaman, J. Sapu- design for a class of nonlinear fractional chaotic sys-
tra, Bernoulli fractional differential equation solution tems. Neural Comput. Appl. 31(4), 1201–1213 (2019)
using adomian decomposition method, in IOP Confer- 140. H. Qu, X. Liu, A numerical method for solving frac-
ence Series: Materials Science and Engineering, vol. tional differential equations by using neural network.
1115, no. 1 (IOP Publishing, 2021), p. 012015 Adv. Math. Phys. 3, 2015 (2015)
125. M.Z. Mohamed, T.M. Elzaki, M.S. Algolam, E.M. 141. A. Jafarian, M. Mokhtarpour, D. Baleanu, Artifi-
Abd Elmohmoud, A.E. Hamza, New modified varia- cial neural network approach for a class of fractional
tional iteration Laplace transform method compares ordinary differential equation. Neural Comput. Appl.
laplace adomian decomposition method for solution 28(4), 765–773 (2017)
time-partial fractional differential equations. J. Appl. 142. S.S. Chaharborj, S.S. Chaharborj, Y. Mahmoudi,
Math. 2021, 1–18 (2021) Study of fractional order integro-differential equations
126. M. Nadeem, J.-H. He, A. Islam, The homotopy per- by using Chebyshev neural network. J. Math. Stat.
turbation method for fractional differential equations: 13(1), 1–13 (2017)
part 1 Mohand transform. Int. J. Numer. Methods Heat 143. M.I. Nouh, Y.A. Azzam, E.A. Abdel-Salam, Modeling
Fluid Flow 1, 1–17 (2021) fractional polytropic gas spheres using artificial neu-
127. A. Atangana, R.T. Alqahtani, New numerical method ral network. Neural Comput. Appl. (2020). https://doi.
and application to Keller–Segel model with fractional org/10.1007/s00521-020-05277-9
order derivative. Chaos Solitons Fract. 116, 14–21 144. M.Ö. Efe, Neural network assisted computationally
(2018) simple PIλ Dμ control of a quadrotor UAV. IEEE
128. R. Garrappa, Numerical solution of fractional differen- Trans. Industr. Inf. 7(2), 354–361 (2011)
tial equations: a survey and a software tutorial. Math- 145. S. Wang, M. Yang, Y. Zhang, J. Li, L. Zou, S. Lu,
ematics 6(2), 16 (2018) B. Liu, J. Yang, Y. Zhang, Detection of left-sided and
129. M. Toufik, A. Atangana, New numerical approximation right-sided hearing loss via fractional Fourier trans-
of fractional derivative with non-local and non-singular form. Entropy 18(5), 1–10 (2016)
kernel: application to chaotic models. Eur. Phys. J. 146. S. Lodhi, M.A. Manzar, M.A.Z. Raja, Fractional neural
Plus 132(10), 1–16 (2017) network models for nonlinear Riccati systems. Neural
130. A. Atangana, S.İ Araz, New numerical method for Comput. Appl. 31, 359–378 (2019)
ordinary differential equations: Newton polynomial. J. 147. M.A.Z. Raja, J.A. Khan, I.M. Qureshi, Solution of frac-
Comput. Appl. Math. 372, 112622 (2020) tional order system of Bagley–Torvik equation using
131. S. Khan, J. Ahmad, I. Naseem, M. Moinuddin, A novel evolutionary computational intelligence. Math. Probl.
fractional gradient-based learning algorithm for recur- Eng. 2011, 1–15 (2011)
rent neural networks. Circuits Syst. Signal Process. 148. Javad Sabouri K, S. Effati, M. Pakdaman, A neural
37(2), 593–612 (2018) network approach for solving a class of fractional opti-
132. J. Wang, Y. Wen, Y. Gou, Z. Ye, H. Chen, Fractional- mal control problems. Neural Process. Lett. 45(1), 59–
order gradient descent learning of BP neural net- 74 (2017)
works with Caputo derivative. Neural Netw. 89, 19–30 149. A. Asgharnia, A. Jamali, R. Shahnazi, A. Maheri, Load
(2017). https://doi.org/10.1016/j.neunet.2017.02.007 mitigation of a class of 5-MW wind turbine with RBF
133. Y. Chen, G. Zhao, A Caputo-type fractional-order gra- neural network based fractional-order PID controller.
dient descent learning of deep BP neural networks, in ISA Trans. 96, 272–286 (2020). https://doi.org/10.
Proceedings of 2019 IEEE 3rd Advanced Information 1016/j.isatra.2019.07.006
Management, Communicates, Electronic and Automa- 150. A. Shaikh, M. AsifJamal, F. Hanif, M. Sadiq AliKhan,
tion Control Conference, IMCEC 2019, no. Imcec S. Inayatullah, Neural minimization methods (NMM)
(2019), pp. 546–550 for solving variable order fractional delay differential
123
equations (FDDEs) with simulated annealing (SA). 165. N. Liu, J. Fei, Adaptive fractional sliding mode con-
PLoS One 14(10), 1–22 (2019) trol of active power filter based on dual RBF neural
151. L. Zhang, Y. Yang, Optimal quasi-synchronization networks. IEEE Access 5, 27 590-27 598 (2017)
of fractional-order memristive neural networks with 166. B. Meng, X. Wang, Adaptive synchronization for
PSOA. Neural Comput. Appl. 32(13), 9667–9682 uncertain delayed fractional-order hopfield neural net-
(2020). https://doi.org/10.1007/s00521-019-04488-z works via fractional-order sliding mode control. Math.
152. A. Mohammadzadeh, E. Kayacan, A novel fractional- Probl. Eng. 2014, 1–17 (2018)
order type-2 fuzzy control method for online frequency 167. H. Bao, J.H. Park, J. Cao, Synchronization of
regulation in ac microgrid. Eng. Appl. Artif. Intell. fractional-order complex-valued neural networks with
90(2018), 103483 (2020). https://doi.org/10.1016/j. time delay. Neural Netw. 81, 16–28 (2016). https://
engappai.2020.103483 doi.org/10.1016/j.neunet.2016.05.003
153. Q. Chang, A. Hu, Y. Yang, L. Li, The optimization of 168. J. Zhang, J. Wu, H. Bao, J. Cao, Synchronization anal-
synchronization control parameters for fractional-order ysis of fractional-order three-neuron BAM neural net-
delayed memristive neural networks using SIWPSO. works with multiple time delays. Appl. Math. Com-
Neural Process. Lett. 51(2), 1541–1556 (2020) put. 339, 441–450 (2018). https://doi.org/10.1016/j.
154. L. Zhang, Y. Yang, F. Wang, Synchronization analysis amc.2018.06.013
of fractional-order neural networks with time-varying 169. T. Ma, J. Zhang, Hybrid synchronization of cou-
delays via discontinuous neuron activations. Neuro- pled fractional-order complex networks. Neurocomput-
computing 275, 40–49 (2018) ing 157, 166–172 (2015). https://doi.org/10.1016/j.
155. H. Liu, S. Li, H. Wang, Y. Huo, J. Luo, Adaptive syn- neucom.2015.01.022
chronization for a class of uncertain fractional-order 170. S. Zhou, H. Li, Z. Zhu, Chaos control and synchro-
neural networks. Entropy 17(10), 7185–7200 (2015) nization in a fractional neuron network system. Chaos
156. S. Luo, S. Li, F. Tajaddodianfar, J. Hu, Adaptive syn- Solitons Fract. 36(4), 973–984 (2008)
chronization of the fractional-order chaotic arch micro- 171. J. Yu, C. Hu, H. Jiang, α-stability and α-
electro-mechanical system via Chebyshev neural net- synchronization for fractional-order neural networks.
work. IEEE Sens. J. 18(9), 3524–3532 (2018) Neural Netw. 35, 82–87 (2012). https://doi.org/10.
157. S. Song, X. Song, I.T. Balsera, Adaptive projective 1016/j.neunet.2012.07.009
synchronization for fractional-order T-S fuzzy neural 172. L. Zhang, Y. Yang, F. Wang, Lag synchronization for
networks with time-delay and uncertain parameters. fractional-order memristive neural networks via period
Optik 129, 140–152 (2017). https://doi.org/10.1016/j. intermittent control. Nonlinear Dyn. 89(1), 367–381
ijleo.2016.10.067 (2017)
158. X. Song, S. Song, I.T. Balsera, Mixed H∞/passive pro- 173. W. Zhang, R. Wu, J. Cao, A. Alsaedi, T. Hayat, Syn-
jective synchronization for fractional-order neural net- chronization of a class of fractional-order neural net-
works with uncertain parameters and delays. Int. J. works with multiple time delays by comparison prin-
Innov. Comput. Inf. Control 13(4), 1273–1288 (2017) ciples. Nonlinear Anal. Model. Control 22(5), 636–645
159. H.B. Bao, J.D. Cao, Projective synchronization (2017)
of fractional-order memristor-based neural networks. 174. L. Chen, R. Wu, J. Cao, J.B. Liu, Stability and
Neural Netw. 63, 1–9 (2015). https://doi.org/10.1016/ synchronization of memristor-based fractional-order
j.neunet.2014.10.007 delayed neural networks. Neural Netw. 71, 37–44
160. J. Yu, C. Hu, H. Jiang, Corrigendum to “Projective (2015). https://doi.org/10.1016/j.neunet.2015.07.012
synchronization for fractional neural networks ’ ’. Neu- 175. M. Zheng, L. Li, H. Peng, J. Xiao, Y. Yang,
ral Netw. 67, 152–154 (2015). https://doi.org/10.1016/ H. Zhao, Finite-time stability and synchronization
j.neunet.2015.02.007 for memristor-based fractional-order Cohen–Grossberg
161. T.C. Lin, C.H. Kuo, H∞ synchronization of uncer- neural network. Eur. Physi. J. B 89(9), 1 (2016)
tain fractional order chaotic systems: adaptive fuzzy 176. M. Zheng, L. Li, H. Peng, J. Xiao, Y. Yang, Y. Zhang,
approach. ISA Trans. 50(4), 548–556 (2011). https:// H. Zhao, Finite-time stability and synchronization of
doi.org/10.1016/j.isatra.2011.06.001 memristor-based fractional-order fuzzy cellular neural
162. J. Chen, C. Li, X. Yang, Global Mittag–Leffler pro- networks. Commun. Nonlinear Sci. Numer. Simul. 59,
jective synchronization of nonidentical fractional-order 272–291 (2018). https://doi.org/10.1016/j.cnsns.2017.
neural networks with delay via sliding mode con- 11.025
trol. Neurocomputing 313, 324–332 (2018). https:// 177. M. Zheng, L. Li, H. Peng, J. Xiao, Y. Yang, H. Zhao,
doi.org/10.1016/j.neucom.2018.06.029 Finite-time projective synchronization of memristor-
163. H. Wu, L. Wang, P. Niu, Y. Wang, Global pro- based delay fractional-order neural networks. Nonlin-
jective synchronization in finite time of nonidentical ear Dyn. 89(4), 2641–2655 (2017)
fractional-order neural networks based on sliding mode 178. L. Zhang, Y. Yang, F. Wang, Projective synchroniza-
control strategy. Neurocomputing 235(January), 264– tion of fractional-order memristive neural networks
273 (2017). https://doi.org/10.1016/j.neucom.2017.01. with switching jumps mismatch. Phys. A 471, 402–415
022 (2017). https://doi.org/10.1016/j.physa.2016.12.030
164. Y. Xi, Y. Yu, S. Zhang, X. Hai, Finite-time robust con- 179. F. Zouari, A. Boulkroune, A. Ibeas, Neural adap-
trol of uncertain fractional-order Hopfield neural net- tive quantized output-feedback control-based syn-
works via sliding mode control. Chin. Phys. B 27(1), chronization of uncertain time-delay incommensurate
010202 (2018) fractional-order chaotic systems with input nonlinear-
123
ities. Neurocomputing 237, 200–225 (2017). https:// 194. B. Meng, Z. Wang, Z. Wang, Adaptive sliding mode
doi.org/10.1016/j.neucom.2016.11.036 control for a class of uncertain nonlinear fractional-
180. Y. Gu, Y. Yu, H. Wang, Synchronization for fractional- order Hopfield neural networks. AIP Adv. 10(1063/1),
order time-delayed memristor-based neural networks 5097374 (2019)
with parameter uncertainty. J. Frankl. Inst. 353(15), 195. S. Song, X. Song, I. Tejado, Projective synchroniza-
3657–3684 (2016). https://doi.org/10.1016/j.jfranklin. tion for two nonidentical time-delayed fractional-order
2016.06.029 T–S fuzzy neural networks based on mixed H∞ /pas-
181. Z. Ding, Y. Shen, L. Wang, Global Mittag–Leffler syn- sive adaptive sliding mode control. Int. J. Mach. Learn.
chronization of fractional-order neural networks with Cybern. 10(5), 799–812 (2019). https://doi.org/10.
discontinuous activations. Neural Netw. 73, 77–85 1007/s13042-017-0761-x
(2016). https://doi.org/10.1016/j.neunet.2015.10.010 196. Y.L. Wang, H. Jahanshahi, S. Bekiros, F. Bezzina,
182. X. Li, J.-A. Fang, W. Zhang, H. Li, Finite-time Y.M. Chu, A.A. Aly, Deep recurrent neural networks
synchronization of fractional-order memristive recur- with finite-time terminal sliding mode control for a
rent neural networks with discontinuous activation chaotic fractional-order financial system with market
functions R. Neurocomputing 316, 284–293 (2018). confidence. Chaos Solitons Fract. 146, 110881 (2021).
https://doi.org/10.1016/j.neucom.2018.08.003 https://doi.org/10.1016/j.chaos.2021.110881
183. W. Zhang, J. Cao, D. Chen, F.E. Alsaadi, Synchroniza- 197. A. Sharafian, A. Sharifi, W. Zhang, Fractional sliding
tion in fractional-order complex-valued delayed neural mode based on RBF neural network observer: applica-
networks. Entropy 20(1), 1–16 (2018) tion to HIV infection mathematical model. Comput.
184. H. Bao, J.H. Park, J. Cao, Adaptive synchronization of Math. Appl. 79(11), 3179–3188 (2020). https://doi.
fractional-order memristor-based neural networks with org/10.1016/j.camwa.2020.01.014
time delay. Nonlinear Dyn. 82(3), 1343–1354 (2015) 198. J. Fei, H. Wang, Experimental investigation of recur-
185. I. Stamova, Global Mittag–Leffler stability and syn- rent neural network fractional-order sliding mode con-
chronization of impulsive fractional-order neural net- trol of active power filter. IEEE Trans. Circuits Syst.
works with time-varying delays. Nonlinear Dyn. 77(4), II Expr. Briefs PP(c), 1–1 (2019)
1251–1260 (2014) 199. Fei, Wang, Recurrent neural network fractional-order
186. I. Stamova, G. Stamov, Mittag–Leffler synchronization sliding mode control of dynamic systems. J. Frankl.
of fractional neural networks with time-varying delays Inst. 357(8), 4574–4591 (2020). https://doi.org/10.
and reaction-diffusion terms using impulsive and linear 1016/j.jfranklin.2020.01.050
controllers. Neural Netw. 96, 22–32 (2017). https:// 200. Z. Han, S. Li, H. Liu, Composite learning sliding mode
doi.org/10.1016/j.neunet.2017.08.009 synchronization of chaotic fractional-order neural net-
187. S. Zhou, X. Lin, H. Li, Chaotic synchronization of a works. J. Adv. Res. (2020). https://doi.org/10.1016/j.
fractional-order system based on washout filter control. jare.2020.04.006
Commun. Nonlinear Sci. Numer. Simul. 16(3), 1533– 201. W. Wang, X. Jia, Z. Wang, X. Luo, L. Li, J. Kurths, M.
1540 (2011). https://doi.org/10.1016/j.cnsns.2010.06. Yuan, Fixed-time synchronization of fractional order
022 memristive MAM neural networks by sliding mode con-
188. X. Zhang, C. Yang, Neural network synchronization trol. Neurocomputing 401, 364–376 (2020)
of fractional-order chaotic systems subject to back- 202. A. Pratap, R. Raja, J. Cao, F.A. Rihan, A.R. Seadawy,
lash nonlinearity. AIP Adv. (2020). https://doi.org/10. Quasi-pinning synchronization and stabilization of
1063/5.0007914 fractional order BAM neural networks with delays and
189. G. Li, C. Sun, Adaptive neural network backstepping discontinuous neuron activations. Chaos Solitons Fract.
control of fractional-order Chua–Hartley chaotic sys- 131, 109491 (2020)
tem. Adv. Differ. Equ. 1, 2019 (2019). https://doi.org/ 203. M.V. Thuan, T.N. Binh, D.C. Huong, Finite-time guar-
10.1186/s13662-019-2099-z anteed cost control of Caputo fractional-order neural
190. Y. Kao, Y. Li, J.H. Park, X. Chen, Mittag–Leffler networks. Asian J. Control 22(2), 696–705 (2020)
synchronization of delayed fractional memristor neu- 204. T. Hu, Z. He, X. Zhang, S. Zhong, Global synchro-
ral networks via adaptive control. IEEE Trans. Neural nization of time-invariant uncertainty fractional-order
Netw. Learn. Syst. 1, 1–6 (2020) neural networks with time delay. Neurocomputing 339,
191. X. Zhang, X. Zhang, D. Li, D. Yang, Adaptive synchro- 45–58 (2019). https://doi.org/10.1016/j.neucom.2019.
nization for a class of fractional order time-delay uncer- 02.020
tain chaotic systems via fuzzy fractional order neural 205. P. Liu, M. Kong, Z. Zeng, Projective synchronization
network. Int. J. Control Autom. Syst. 17(106112016), analysis of fractional-order neural networks with mixed
1–12 (2019) time delays. IEEE Trans. Cybern. 1, 1–11 (2020)
192. W. Cheng, A. Wu, J.E. Zhang, B. Li, Adaptive control 206. H. Yan, Y. Qiao, L. Duan, L. Zhang, Global Mittag-
of Mittag–Leffler stabilization and synchronization for Leffler stabilization of fractional-order BAM neural
delayed fractional-order BAM neural networks. Adv. networks with linear state feedback controllers. Math.
Differ. Equ. 1, 2019 (2019). https://doi.org/10.1186/ Probl. Eng. 2020, 1–14 (2020)
s13662-019-2279-x 207. Z. Yang, J. Li, Y. Niu, Finite-time stabilization of
193. J. He, F. Chen, T. Lei, Q. Bi, Global adaptive matrix- fractional-order delayed bidirectional associative mem-
projective synchronization of delayed fractional-order ory neural networks. ScienceAsia 45(6), 589–596 (2019)
competitive neural network with different time scales. 208. Y. Fan, X. Huang, Z. Wang, J. Xia, H. Shen, Quan-
Neural Comput. Appl. 32(16), 12 813-12 826 (2020). tized control for synchronization of delayed fractional-
https://doi.org/10.1007/s00521-020-04728-7 order memristive neural networks. Neural Process.
123
Lett. 52(1), 403–419 (2020). https://doi.org/10.1007/ 222. M. Syed Ali, M. Hymavathi, S. Senan, V. Shekher, S.
s11063-020-10259-y Arik, Global asymptotic synchronization of impulsive
209. Y. Gu, Y. Yu, H. Wang, Projective synchronization for fractional-order complex-valued memristor-based neu-
fractional-order memristor-based neural networks with ral networks with time varying delays. Commun. Non-
time delays. Neural Comput. Appl. 31(10), 6039–6054 linear Sci. Numer. Simul. 78, 104869 (2019). https://
(2019). https://doi.org/10.1007/s00521-018-3391-7 doi.org/10.1016/j.cnsns.2019.104869
210. J. Jia, X. Huang, Y. Li, J. Cao, A. Alsaedi, Global sta- 223. G. Rajchakit, A. Pratap, R. Raja, J. Cao, J. Alzabut,
bilization of fractional-order memristor-based neural C. Huang, Hybrid control scheme for projective lag
networks with time delay. IEEE Trans. Neural Netw. synchronization of Riemann–Liouville sense fractional
Learn. Syst. 31(3), 997–1009 (2020) order memristive BAM neural networks with mixed
211. C. Chen, Z. Ding, S. Li, L. Wang, Synchronization delays. Mathematics 7(8), 759 (2019)
of fractional-order memristive neural networks with 224. A. Pratap, R. Raja, R.P. Agarwal, J. Cao, Stability
time delays, in Proceedings—2019 Chinese Automation analysis and robust synchronization of fractional-order
Congress. CAC 2019 (2019), pp. 2754–2759 competitive neural networks with different time scales
212. L. Chen, T. Huang, J.A. Tenreiro Machado, A.M. and impulsive perturbations. Int. J. Adapt. Control
Lopes, Y. Chai, R. Wu, Delay-dependent criterion Signal Process. 33(11), 1635–1660 (2019)
for asymptotic stability of a class of fractional-order 225. Z. Yang, J. Zhang, Global stabilization of fractional-
memristive neural networks with time-varying delays. order bidirectional associative memory neural networks
Neural Netw. 118, 289–299 (2019). https://doi.org/10. with mixed time delays via adaptive feedback control.
1016/j.neunet.2019.07.006 Int. J. Comput. Math. 0(0), 1–17 (2019). https://doi.
213. J. Xiao, J. Cao, J. Cheng, S. Zhong, S. Wen, Novel org/10.1080/00207160.2019.1677897
methods to finite-time Mittag–Leffler synchronization 226. P. Anbalagan, R. Ramachandran, J. Cao, G. Rajchakit,
problem of fractional-order quaternion-valued neural C.P. Lim, Global Robust synchronization of fractional
networks. Inf. Sci. 526, 221–244 (2020). https://doi. order complex valued neural networks with mixed time
org/10.1016/j.ins.2020.03.101 varying delays and impulses. Int. J. Control Autom.
214. J. Xiao, S. Wen, X. Yang, S. Zhong, New Syst. 17(2), 509–520 (2019)
approach to global Mittag–Leffler synchronization 227. P. Wan, J. Jian, Impulsive stabilization and synchro-
problem of fractional-order quaternion-valued BAM nization of fractional-order complex-valued neural net-
neural networks based on a new inequality. Neural works. Neural Process. Lett. 50(3), 2201–2218 (2019).
Netw. 122, 320–337 (2020). https://doi.org/10.1016/ https://doi.org/10.1007/s11063-019-10002-2
j.neunet.2019.10.017 228. L. Zhang, Y. Yang, Finite time impulsive synchroniza-
215. G. Rajchakit, P. Chanthorn, P. Kaewmesri, R. Sri- tion of fractional order memristive BAM neural net-
raman, C.P. Lim, Global Mittag–Leffler stability and works. Neurocomputing 384, 213–224 (2020). https://
stabilization analysis of fractional-order quaternion- doi.org/10.1016/j.neucom.2019.12.056
valued memristive neural networks. Mathematics 8(3), 229. L. Zhang, Stability analysis of fractional order mem-
422 (2020) ristive discontinuous neural networks with partial state
216. H.L. Li, L. Zhang, C. Hu, H. Jiang, J. Cao, control. Phys. A 531(18), 121756 (2019). https://doi.
Global Mittag–Leffler synchronization of fractional- org/10.1016/j.physa.2019.121756
order delayed quaternion-valued neural networks: 230. J. Cao, G. Stamov, I. Stamova, S. Simeonov, Almost
direct quaternion approach. Appl. Math. Comput. 373, periodicity in impulsive fractional-order reaction–
125020 (2020) diffusion neural networks with time-varying delays.
217. S.A. Karthick, R. Sakthivel, Y.K. Ma, S. Mohanapriya, IEEE Trans. Cybern. 3, 1–11 (2020)
A. Leelamani, Disturbance rejection of fractional-order 231. A. Pratap, R. Raja, J. Alzabut, J. Cao, G. Rajchakit,
T–S fuzzy neural networks based on quantized dynamic C. Huang, Mittag–Leffler stability and adaptive impul-
output feedback controller. Appl. Math. Comput. 361, sive synchronization of fractional order neural networks
846–857 (2019) in quaternion field. Math. Methods Appl. Sci. 43(10),
218. A. Pratap, R. Raja, J. Cao, J. Alzabut, C. Huang, 6223–6253 (2020)
Finite-time synchronization criterion of graph theory 232. Z.B. Wu, Y.Z. Zou, N.J. Huang, A system of fractional-
perspective fractional-order coupled discontinuous neu- order interval projection neural networks. J. Comput.
ral networks. Adv. Differ. Equ. 1, 2020 (2020). https:// Appl. Math. 294, 389–402 (2016). https://doi.org/10.
doi.org/10.1186/s13662-020-02551-x 1016/j.cam.2015.09.007
219. C. Wang, Q. Yang, Y. Zhuo, R. Li, Synchronization 233. L. Zhang, Q. Song, Z. Zhao, Stability analysis of
analysis of a fractional-order non-autonomous neural fractional-order complex-valued neural networks with
network with time delay. Phys. A 549, 124176 (2020). both leakage and discrete delays. Appl. Math. Com-
https://doi.org/10.1016/j.physa.2020.124176 put. 298, 296–309 (2017). https://doi.org/10.1016/j.
220. Y. Gu, H. Wang, Y. Yu, Stability and synchronization amc.2016.11.027
for Riemann-Liouville fractional-order time-delayed 234. S. Tyagi, S. Abbas, M. Hafayed, Global Mittag–Leffler
inertial neural networks. Neurocomputing 340, 270– stability of complex valued fractional-order neural net-
280 (2019). https://doi.org/10.1016/j.neucom.2019.03. work with discrete and distributed delays. Rendiconti
005 del Circolo Matematico di Palermo 65(3), 485–505
221. Y. Gu, H. Wang, Y. Yu, Synchronization for fractional- (2016)
order discrete-time neural networks with time delays.
Appl. Math. Comput. 372, 124995 (2020)
123
235. Y. Cao, C. Bai, Existence and stability analysis of frac- trary neurons. Commun. Nonlinear Sci. Numer. Simul.
tional order BAM neural networks with a time delay. 57, 1–13 (2018). https://doi.org/10.1016/j.cnsns.2017.
Appl. Math. 06(12), 2057–2068 (2015) 09.005
236. X. Yang, Q. Song, Y. Liu, Z. Zhao, Uniform stability 251. E. Kaslik, I.R. Rǎdulescu, Dynamics of complex-valued
analysis of fractional-order bam neural networks with fractional-order neural networks. Neural Netw. 89,
delays in the leakage terms. Abstr. Appl. Anal. 2014, 39–49 (2017). https://doi.org/10.1016/j.neunet.2017.
1–15 (2014) 02.011
237. Y. Yang, Y. He, Y. Wang, M. Wu, Stability analysis 252. H. Wu, X. Zhang, S. Xue, P. Niu, Quasi-uniform sta-
of fractional-order neural networks: an LMI approach. bility of Caputo-type fractional-order neural networks
Neurocomputing 285, 82–93 (2018). https://doi.org/ with mixed delay. Int. J. Mach. Learn. Cybern. 8(5),
10.1016/j.neucom.2018.01.036 1501–1511 (2017)
238. Y. Zhao, Y. Cai, G. Fan, Dynamical behavior for 253. N. Sene, Fractional input stability and its application
fractional-order shunting inhibitory cellular neural net- to neural network. Discrete Cont. Dyn. Syst. 13, 853–
works. J. Nonlinear Sci. Appl. 09(06), 4589–4599 865 (2018)
(2018) 254. R. Rakkiyappan, R. Sivaranjani, G. Velmurugan, J.
239. R. Rakkiyappan, G. Velmurugan, J. Cao, Stability Cao, Analysis of global O(t-α) stability and global
analysis of memristor-based fractional-order neural asymptotical periodicity for a class of fractional-order
networks with different memductance functions. Cogn. complex-valued neural networks with time varying
Neurodyn. 9(2), 145–177 (2015) delays. Neural Netw. 77, 51–69 (2016). https://doi.
240. S. Zhang, Y. Yu, H. Wang, Mittag–Leffler stability org/10.1016/j.neunet.2016.01.007
of fractional-order Hopfield neural networks. Nonlin- 255. H. Zhang, R. Ye, J. Cao, A. Alsaedi, Existence and
ear Anal. Hybrid Syst 16(11371049), 104–121 (2015). globally asymptotic stability of equilibrium solution for
https://doi.org/10.1016/j.nahs.2014.10.001 fractional-order hybrid BAM neural networks with dis-
241. C. Song, J. Cao, Dynamics in fractional-order neu- tributed delays and impulses. Complexity 2017, 1–15
ral networks. Neurocomputing 142, 494–498 (2014). (2017)
https://doi.org/10.1016/j.neucom.2014.03.047 256. H. Wu, X. Zhang, S. Xue, L. Wang, Y. Wang, LMI con-
242. L. Chen, Y. Chai, R. Wu, T. Ma, H. Zhai, Dynamic ditions to global Mittag–Leffler stability of fractional-
analysis of a class of fractional-order neural networks order neural networks with impulses. Neurocomput-
with delay. Neurocomputing 111, 190–194 (2013). ing 193, 148–154 (2016). https://doi.org/10.1016/j.
https://doi.org/10.1016/j.neucom.2012.11.034 neucom.2016.02.002
243. X. Ding, J. Cao, X. Zhao, F.E. Alsaadi, Finite-time 257. G. Velmurugan, R. Rakkiyappan, V. Vembarasan, J.
stability of fractional-order complex-valued neural net- Cao, A. Alsaedi, Dissipativity and stability analysis of
works with time delays. Neural Process. Lett. 46(2), fractional-order complex-valued neural networks with
561–580 (2017) time delay. Neural Netw. 86, 42–53 (2017). https://
244. L. Wang, Q. Song, Y. Liu, Z. Zhao, F.E. Alsaadi, doi.org/10.1016/j.neunet.2016.10.010
Finite-time stability analysis of fractional-order 258. G. Stamov, I. Stamova, Impulsive fractional-order neu-
complex-valued memristor-based neural networks with ral networks with time-varying delays: almost periodic
both leakage and time-varying delays. Neurocomput- solutions. Neural Comput. Appl. 28(11), 3307–3316
ing 245, 86–101 (2017). https://doi.org/10.1016/j. (2017)
neucom.2017.03.042 259. F. Wang, Y. Yang, M. Hu, Asymptotic stability of
245. Y. Ke, C. Miao, Stability analysis of fractional-order delayed fractional-order neural networks with impul-
Cohen–Grossberg neural networks with time delay. Int. sive effects. Neurocomputing 154, 239–244 (2015).
J. Comput. Math. 92(6), 1102–1113 (2015) https://doi.org/10.1016/j.neucom.2014.11.068
246. L. Chen, C. Liu, R. Wu, Y. He, Y. Chai, Finite-time 260. R.P. Agarwal, S. Hristova, D. O’regan, Lyapunov func-
stability criteria for a class of fractional-order neu- tions to Caputo reaction–diffusion fractional neural
ral networks with delay. Neural Comput. Appl. 27(3), networks with time-varying delays. J. Math. Comput.
549–556 (2015) Sci. 18(3), 328–345 (2018)
247. A. Alofi, J. Cao, A. Elaiw, A. Al-Mazrooei, Delay- 261. S. Zhang, Y. Yu, Q. Wang, Stability analysis of
dependent stability criterion of caputo fractional neu- fractional-order Hopfield neural networks with dis-
ral networks with distributed delay. Discrete Dyn. Nat. continuous activation functions. Neurocomputing 171,
Soc. 1, 1–15 (2014) 1075–1084 (2016). https://doi.org/10.1016/j.neucom.
248. Z. Ding, Z. Zeng, L. Wang, Robust finite-time stabi- 2015.07.077
lization of fractional-order neural networks with dis- 262. B. Chen, J. Chen, Global asymptotical ω-periodicity
continuous and continuous activation functions under of a fractional-order non-autonomous neural networks.
uncertainty. IEEE Trans. Neural Netw. Learn. Syst. Neural Netw. 68, 78–88 (2015). https://doi.org/10.
29(5), 1477–1490 (2018) 1016/j.neunet.2015.04.006
249. C. Huang, Y. Meng, J. Cao, A. Alsaedi, F.E. Alsaadi, 263. S. Mehdi, A. Pahnehkolaei, A. Alfi, J.A.T. Machado,
New bifurcation results for fractional BAM neural net- Dynamic stability analysis of fractional order leaky
work with leakage delay. Chaos Solitons Fract. 100, integrator echo state neural networks. Commun. Non-
31–44 (2017) linear Sci. Numer. Simul. 47, 328–337 (2017). https://
250. C. Huang, J. Cao, M. Xiao, A. Alsaedi, T. Hayat, doi.org/10.1016/j.cnsns.2016.11.013
Effects of time delays on stability and Hopf bifurca- 264. H. Zhang, R. Ye, J. Cao, A. Alsaedi, Delay-independent
tion in a fractional ring-structured network with arbi- stability of Riemann–Liouville fractional neutral-type
123
delayed neural networks. Neural Process. Lett. 47(2), 278. G. Nagamani, M. Shafiya, G. Soundararajan, M.
427–442 (2017) Prakash, Robust state estimation for fractional-order
265. S. Liang, R. Wu, L. Chen, Comparison principles and delayed BAM neural networks via LMI approach. J.
stability of nonlinear fractional-order cellular neural Frankl. Inst. 357(8), 4964–4982 (2020)
networks with multiple time delays. Neurocomput- 279. W. Liu, M. Jiang, M. Yan, Stability analysis
ing 168, 618–625 (2015). https://doi.org/10.1016/j. of memristor-based time-delay fractional-order neu-
neucom.2015.05.063 ral networks. Neurocomputing 323, 117–127 (2019).
266. H. Wei, R. Li, C. Chen, Z. Tu, Stability Analysis of https://doi.org/10.1016/j.neucom.2018.09.073
fractional order complex-valued memristive neural net- 280. M. Syed Ali, G. Narayanan, S. Sevgen, V. Shekher, S.
works with time delays. Neural Process. Lett. 45(2), Arik, Global stability analysis of fractional-order fuzzy
379–399 (2017) BAM neural networks with time delay and impulsive
267. A. Wu, L. Liu, T. Huang, Z. Zeng, Mittag–Leffler sta- effects. Commun. Nonlinear Sci. Numer. Simul. 78,
bility of fractional-order neural networks in the pres- 104853 (2019). https://doi.org/10.1016/j.cnsns.2019.
ence of generalized piecewise constant arguments. Neu- 104853
ral Netw. 85, 118–127 (2017). https://doi.org/10.1016/ 281. C. Huang, X. Zhao, X. Wang, Z. Wang, M. Xiao,
j.neunet.2016.10.002 J. Cao, Disparate delays-induced bifurcations in
268. R. Li, J. Cao, A. Alsaedi, F. Alsaadi, Stability analysis a fractional-order neural network. J. Frankl. Inst.
of fractional-order delayed neural networks. Nonlinear 356(5), 2825–2846 (2019). https://doi.org/10.1016/j.
Anal. Model. Control 22(4), 505–520 (2017) jfranklin.2018.11.027
269. B. Chen, J. Chen, Global O(t α) stability and 282. C. Xu, M. Liao, P. Li, J. Yan, Influence of time delay
global asymptotical periodicity for a non-autonomous on bifurcation in fractional order BAM neural networks
fractional-order neural networks with time-varying with four delays. IEEE Access 7, 70 955-70 965 (2019)
delays. Neural Netw. 73, 47–57 (2016). https://doi. 283. P. Wan, D. Sun, M. Zhao, H. Zhao, Monostabil-
org/10.1016/j.neunet.2015.09.007 ity and multistability for almost-periodic solutions
270. H. Zhang, M. Ye, R. Ye, J. Cao, Synchronization stabil- of fractional-order neural networks with unsaturating
ity of Riemann–Liouville fractional delay-coupled com- piecewise linear activation functions. IEEE Tran. Neu-
plex neural networks. Phys. A 508, 155–165 (2018). ral Netw. Learn. Syst. 1, 1–15 (2020)
https://doi.org/10.1016/j.physa.2018.05.060 284. L. Wan, X. Zhan, H. Gao, Q. Yang, T. Han, M. Ye,
271. M. Tuz, G.A. Suroglu, Anti-periodic solutions for Multiple asymptotical stability analysis for fractional-
fractional-order bidirectional associative memory neu- order neural networks with time delays. Int. J. Syst.
ral networks with delays. Therm. Sci. 23, S2169–S2177 Sci. 50(10), 2063–2076 (2019). https://doi.org/10.
(2019) 1080/00207721.2019.1646836
272. X. Yao, S. Zhong, T. Hu, H. Cheng, D. Zhang, 285. G. Rajchakit, P. Chanthorn, M. Niezabitowski, R.
Uniformly stable and attractive of fractional-order Raja, D. Baleanu, A. Pratap, Impulsive effects on
memristor-based neural networks with multiple delays. stability and passivity analysis of memristor-based
Appl. Math. Comput. 347, 392–403 (2019). https:// fractional-order competitive neural networks. Neuro-
doi.org/10.1016/j.amc.2018.11.028 computing 417, 290–301 (2020). https://doi.org/10.
273. M. Syed Ali, G. Narayanan, V. Shekher, H. Alsu- 1016/j.neucom.2020.07.036
lami, T. Saeed, Dynamic stability analysis of stochas- 286. G.A. Anastassiou, Fractional neural network approx-
tic fractional-order memristor fuzzy BAM neural net- imation. Comput. Math. Appl. 64(6), 1655–1676
works with delay and leakage terms. Appl. Math. Com- (2012). https://doi.org/10.1016/j.camwa.2012.01.019
put. 369, 124896 (2020). https://doi.org/10.1016/j. 287. R.W. Ibrahim, The fractional differential polynomial
amc.2019.124896 neural network for approximation of functions. Entropy
274. J. Alzabut, S. Tyagi, S. Abbas, Discrete fractional- 15(10), 4188–4198 (2013)
order BAM neural networks with leakage delay: exis- 288. S. Lu, X. Wang, Observer-based command filtered
tence and stability results. Asian J. Control 22(1), 143– adaptive neural network tracking control for fractional-
155 (2020) order chaotic PMSM. IEEE Access 7, 88 777-88 788
275. A. Jmal, A. Ben Makhlouf, A.M. Nagy, O. Nai- (2019)
far, Finite-time stability for Caputo–Katugampola 289. S. Lu, X. Wang, Y. Li, Adaptive neural network control
fractional-order time-delayed neural networks. Neural for fractional-order PMSM with time delay based on
Process. Lett. 50(1), 607–621 (2019). https://doi.org/ command filtered backstepping. AIP Adv. 10(1063/1),
10.1007/s11063-019-10060-6 5094574 (2019)
276. X. You, Q. Song, Z. Zhao, Existence and finite-time sta- 290. D. Baleanu, Z.B. Güvenç, J.T. Machado (eds.), New
bility of discrete fractional-order complex-valued neu- Trends in Nanotechnology and Fractional Calculus
ral networks with time delays. Neural Netw. 123, 248– Applications (Springer, New York, 2010), p. C397
260 (2020). https://doi.org/10.1016/j.neunet.2019.12. 291. Y. Fan, X. Huang, Z. Wang, Y. Li, Nonlinear dynamics
012 and chaos in a simplified memristor-based fractional-
277. M. Syed Ali, G. Narayanan, Z. Orman, V. Shekher, S. order neural network with discontinuous memduc-
Arik, Finite time stability analysis of fractional-order tance function. Nonlinear Dyn. 93(2), 611–627 (2018).
complex-valued memristive neural networks with pro- https://doi.org/10.1007/s11071-018-4213-2
portional delays. Neural Process. Lett. 51(1), 407–426 292. A.H. Tian, C.B. Fu, X.Y. Su, H.T. Yau, Lathe tool
(2020). https://doi.org/10.1007/s11063-019-10097-7 chatter vibration diagnostic using general regression
neural network based on Chua’s circuit and fractional-
123
order Lorenz master/slave chaotic system. J. Low Freq. 304. W. Li, X. Gao, R. Li, Dissipativity and synchronization
Noise Vib. Act. Control 38(3–4), 953–966 (2019) control of fractional-order memristive neural networks
293. L. Zhang, Y. Yang, Stability analysis of fractional order with reaction–diffusion terms. Math. Methods Appl.
Hopfield neural networks with optimal discontinuous Sci. 42(18), 7494–7505 (2019)
control. Neural Process. Lett. 50(1), 581–593 (2019). 305. D. Sierociuk, I. Petráš, Modeling of heat transfer pro-
https://doi.org/10.1007/s11063-019-10054-4 cess by using discrete fractional-order neural networks,
294. L. Chen, H. Yin, T. Huang, L. Yuan, S. Zheng, L. in 2011 16th International Conference on Methods and
Yin, Chaos in fractional-order discrete neural net- Models in Automation and Robotics, MMAR 2011.
works with application to image encryption. Neural IEEE, pp. 146–150 (2011)
Netw. 125, 174–184 (2020). https://doi.org/10.1016/ 306. Y. Zhang, Y. Sun, P. Phillips, G. Liu, X. Zhou, S.
j.neunet.2020.02.008 Wang, A multilayer perceptron based smart patho-
295. X. Wang, Y. Su, C. Luo, C. Wang, A novel image logical brain detection system by fractional Fourier
encryption algorithm based on fractional order 5D entropy. J. Med. Syst. (2016). https://doi.org/10.1007/
cellular neural network and Fisher–Yates scrambling. s10916-016-0525-2
PLoS ONE 15(7), 1–18 (2020). https://doi.org/10. 307. E.N. Mobarez, A. Sarhan, M. Ashry, Fractional order
1371/journal.pone.0236015 PID based on a single artificial neural network algo-
296. A. Boroomand, M.B. Menhaj, ON-line nonlinear sys- rithm for fixed wing uavs, in ICENCO 2019—2019 15th
tems identification of coupled tanks via fractional dif- International Computer Engineering Conference: Uti-
ferential neural networks, in 2009 Chinese Control and lizing Machine Intelligence for a Better World, (1), pp.
Decision Conference, pp. 2185–2189 (2009) 168–174 (2019)
297. H. Bao, J. Cao, J. Kurths, State estimation of 308. X.Z. Liu, Z.T. Li, K.N. Wu, Boundary Mittag–Leffler
fractional-order delayed memristive neural networks. stabilization of fractional reaction–diffusion cellular
Nonlinear Dyn. 94(2), 1215–1225 (2018). https://doi. neural networks. Neural Netw. 132, 269–280 (2020).
org/10.1007/s11071-018-4419-3 https://doi.org/10.1016/j.neunet.2020.09.009
298. R. Li, X. Gao, J. Cao, Non-fragile state estimation for 309. K. Ding, Q. Zhu, Impulsive method to reliable sampled-
delayed fractional-order memristive neural networks. data control for uncertain fractional-order memris-
Appl. Math. Comput. 340, 221–233 (2019). https:// tive neural networks with stochastic sensor faults and
doi.org/10.1016/j.amc.2018.08.031 its applications. Nonlinear Dyn. 100(3), 2595–2608
299. J. Fei, C. Lu, Adaptive fractional order sliding mode (2020). https://doi.org/10.1007/s11071-020-05670-y
controller with neural estimator. J. Frankl. Inst. 310. L. Sánchez, J. Otero, D. Anseán, I. Couso, Health
355(5), 2369–2391 (2018). https://doi.org/10.1016/j. assessment of LFP automotive batteries using a
jfranklin.2018.01.006 fractional-order neural network. Neurocomputing 391,
300. H. Bao, J.H. Park, J. Cao, Non-fragile state estimation 345–354 (2020)
for fractional-order delayed memristive BAM neural 311. M. Roohi, C. Zhang, Y. Chen, Adaptive model-free
networks. Neural Netw. 119, 190–199 (2019). https:// synchronization of different fractional-order neural net-
doi.org/10.1016/j.neunet.2019.08.003 works with an application in cryptography. Nonlin-
301. R. Li, X. Gao, J. Cao, Quasi-state estimation and ear Dyn. 100(4), 3979–4001 (2020). https://doi.org/
quasi-synchronization control of quaternion-valued 10.1007/s11071-020-05719-y
fractional-order fuzzy memristive neural networks: 312. M. Blasik, Numerical method for the one phase 1D
vector ordering approach. Appl. Math. Comput. fractional Stefan problem supported by an artificial
362(2018), 124572 (2019). https://doi.org/10.1016/j. neural network. (2), 1–20. (2019). arXiv:1909.13638
amc.2019.124572
302. B. Hu, Q. Song, Z. Zhao, Robust state estima-
tion for fractional-order complex-valued delayed neural
networks with interval parameter uncertainties: LMI
approach. Appl. Math. Comput. 373, 125033 (2020).
https://doi.org/10.1016/j.amc.2020.125033
303. M. Hui, C. Wei, J. Zhang, H. Ho-Ching Iu, N. Luo,
R. Yao, L. Bai, Finite-time projective synchronization
of fractional-order memristive neural networks with
mixed time-varying delays. Complexity 2020, 1–16
(2020)
123

Artificial Neural Networks: A Practical Review of Applications Involving Fractional Calculus

Uploaded by

Copyright:

Available Formats

Artificial Neural Networks: A Practical Review of Applications Involving Fractional Calculus

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Artificial Neural Networks: A Practical Review of Applications Involving Fractional Calculus

Uploaded by

Copyright:

Available Formats

THE EUROPEAN

Eur. Phys. J. Spec. Top. (2022) 231:2059–2095

Artiﬁcial neural networks: a practical review of

1 Introduction for describing the dynamical behavior of the ANNs neu-

The interest in applying and solving fractional dif-

[22] Jacobi orthogonal FANN

5 Synchronization and stabilization of ANN

Table 4 ANN involved with FC to reach stability

Only one work related to the FC on ANN with the heat

ANN involved with FC: applications in medicine

ANN involved with FC: applications in image

You might also like