Data protection based neural cryptography and deoxyribonucleic acid

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 12, No. 3, June 2022, pp. 2756~2764
ISSN: 2088-8708, DOI: 10.11591/ijece.v12i3.pp2756-2764  2756
Journal homepage: http://ijece.iaescore.com
Data protection based neural cryptography and
deoxyribonucleic acid
Sahar Adill Kadum, Ali Yakoob Al-Sultan, Najlaa Adnan Hadie
Department of Computer Science, College of Science for Women, Babylon University, Hillah, Iraq
Article Info ABSTRACT
Article history:
Received Nov 20, 2020
Revised Jan 7, 2022
Accepted Jan 19, 2022
The need to a robust and effective methods for secure data transferring
makes the more credible. Two disciplines for data encryption presented in
this paper: machine learning and deoxyribonucleic acid (DNA) to achieve
the above goal and following common goals: prevent unauthorized access
and eavesdropper. They used as powerful tool in cryptography. This paper
grounded first on a two modified Hebbian neural network (MHNN) as a
machine learning tool for message encryption in an unsupervised method.
These two modified Hebbian neural nets classified as a: learning neural net
(LNN) for generating optimal key ciphering and ciphering neural net CNN)
for coding the plaintext using the LNN keys. The second granulation using
DNA nucleated to increase data confusion and compression. Exploiting the
DNA computing operations to upgrade data transmission security over the
open nets. The results approved that the method is effective in protect the
transferring data in a secure manner in less time
Keywords:
Bio-molecular
DNA computing
DNA sequence
Hebbian network
Neural cryptography
This is an open access article under the CC BY-SA license.
Corresponding Author:
Sahar Adill Kadum
Department of Computer Science, College of Science for Women, Babylon University
Hillah, Iraq
Email: dr.sahar.adill@gmail.com
1. INTRODUCTION
Security concerns the safety of the network and the transmission of data. Data security is the most
critical component of the secure transmission of data over the network. Today, however, achieving maximum
data protection is a difficult problem for data communications over open networks. There is more than one
technology used to protect data, such as encryption and information concealment. Encryption codes a
message that cannot be read by an accidental individual. Whereas, hiding information is a technique used to
conceal a hidden message so that it cannot be identified by accidental users [1], [2]. Three goal purposes
from using cryptography: first, to achieve confidentiality. Second, authentic the sender and thirdly, the
integrity. The modern cryptography concentrated on Kirchhoff principle: assume that the attacker known all
or some details of encryption method usage, and the working procedure, except key data piece. Generally, the
cryptography can be classified to two classes: symmetric and asymmetric systems. The symmetry system in
turn classified into two ciphering types: block cipher system and stream cipher system [3] as shown in
Figure 1.
Effective coding techniques are needed to achieve an effective data protection because encryption
method can be down and not robust enough to offer successful data protection. Hence, unauthorized user or
intruders can access the information for various harmful purposes. This leads to the need for adopting
additional supported techniques for cryptography methods to overcome any falls or gaps resulted from
weakness methods such as neural network encryption module concept and deoxyribonucleic acid (DNA)
computing. They incorporated with cryptography and information hiding as a new hope for a robust

Int J Elec & Comp Eng ISSN: 2088-8708 
Data protection based neural cryptography and deoxyribonucleic acid (Sahar Adill Kadum)
2757
cryptographic techniques and unbreakable algorithms [2]. Neural networks have several properties, the most
important of them is the generalization capability and parallel implementation. The special property of neural
networks is confusion caused from non-linearity structure of the net, this confusion property is preferable for
cipher design. Many types of artificial neural networks (ANNs) have built each ANN has own style design
and method [4], one of these types is the Hebbian neural network (HNN). It is a computational tool inspired
by biological nervous system. HNN is a network composed of arrays of artificial neurons linked together
with various connection weights, its rule is a rule of learning that explains how the neuronal connection is
affected by neuronal activities (synaptic plasticity). Therefore, it provides an algorithm for modifying the
neuronal network relationship weight. The Hebb rule offers a simplified model based on physiology to
simulate synaptic plasticity's activity-dependent characteristics and has been widely used in the artificial
neural network field [5]–[9].
Figure 1. Encryption classes [3]
While, DNA computing is an emerging computing branch that utilizes DNA biochemistry. In this
sphere, research and development concerns theory, experiments, and DNA computing applications. The
originally field began with Len Adleman's demonstration of a computer application in 1994, it has now been
extended to many other avenues, such as the advancement of storage technologies, nanoscale imaging
modalities. synthetic controllers and reaction networks [10], [11]. Massive parallelism is the promise of DNA
computing: with a given setup and enough DNA through parallel search once can theoretically solve huge
problems. This could be much faster than a traditional computer, for which large quantities of hardware will
be needed for massive parallelism, not just more DNA [10], [12].
Several research papers have been prepared in the field of cryptography dealing with neural nets and
DNA with various encryption methods and techniques, such as Kulkarni presents a new proposal schedule
for XOR design is considered. they replace the normal XOR function with neural network XOR function in
the design process, the motivation for this is ability of neural networks to perform complex mapping function
to generate key ciphering [13]. Jagtap et al. [14] presents a cryptography based on artificial neural network.
The used artificial neural network (ANN) has many features such as learning, generation, less data
requirement, fast computation, and simple deployment. Mohammed [15] proposed a symmetric cryptography
coupled with a multilevel chaotic neural network (NN), the encryption algorithm process the data as a block.
The proposed algorithm proved the efficiency of execution time in encrypt/decrypt long messages by short
time and small memory, the system uses secret keys with array of keys (weights) that change at each
iteration. Roy et al. [16] for text messages using a time-varying delayed hopfield neural network and a
cryptographic posterior DNA, a new encryption model was proposed. To produce a binary sequence that is
later transferred to a permutation function, the chaotic neural network used here is used and the first level
encryption key is generated. Dixit et al [17] proposed a DNA based cryptography to achieve cryptographic
strength using least significant bits (LSB) method. Applying binary index compression technique to reduces
data up to 50% to improve payload capacity. Output is a sequences of DNA nucleotides. Basu et al. [18]
presents a system of encryption based on the central dogma of molecular biology for encryption/decryption
algorithms. The bidirectional associative memory neural network (BAMNN) has been used for key
generation in order to saving memory space by restoring and regenerating key sets in a recurrent process.
Namasudra et al. [19] proposed a DNA based data encryption scheme for cloud computing era. Generating
key of 1024-bit is depend on DNA computing, user’s attributes and media access control (MAC) address, and
decimal encoding rule, American standard code for information interchange (ASCII values), DNA bases and
complementary rule are to protect against many security attacks. Li et al. [20] proposed a new method of data
protection and user authentication using DNA QR coding to be very fast, reliable and less predictable. Any
message is translated to a DNA sequence and converted to a QR code called a DNA QR code. Singh and
Naidu [21] proposes a method to authenticate the user using DNA sequence at first level and secondly secure
the data using DNA sequence and Armstrong number. Malathi et al. [22] proposed a modified DNA insertion

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 12, No. 3, June 2022: 2756-2764
2758
algorithm due to its low cracking probability. Confidential information, such as confidential messages and
document images, is concealed inside the DNA sequence. Measurement output is performed using cracking
chance, bits per nucleotide (BPN), payload and power. Indrasena et al. [23] presents a bio-inspired
cryptographic DNA system. The scheme is simulating the genetic encoding procedures (transcription and
translation), i.e., the central dogma of molecular biology (CMDB) using BAMNN and whale optimization
algorithm (WOA) as a highest fit weight vector. Volna et al. [24] presents new direction of cryptography
based neural networks, where the cryptographic scheme is generated automatically. The proposal evolving
neural network architecture called Spectrum-diverse unified neuro evolution architecture to achieve
automatic encryption and decryption subsequently using adversarial training. The main purpose behind this
paper is to apply a new encryption method based on two modified HNN for encryption phase and DNA
computing operations for another coding type and compression to increase the concept of data confusion and
confidential secrecy. The paper organized according to the following topics: proposed method, modified
Hebbian neural network (MHNN) structure, research method, results, and conclusion
2. PROPOSED METHOD
The proposed system adopt hybrid securing method for transferring data in an open net by implying
two contribution methods; The first one, adopt a new ciphering method using two MHNN. The second,
exploit the DNA computing operations to compress the data before transferring process and increase the
probability of cracking. Figure 2 illustrates the general structure of the proposed system.
Figure 2. The proposal general structure
2.1. Modified Hebbian neural network (MHNN)
The suggested system designated by adopts the Hebbian neural network infrastructure [5]–[9], but
with several modifications has been done on HNN to the behalf of the proposed system. These modifications
include: i) the number of input nodes and output nodes are equally; ii) the learning process is an unsupervised
approach; iii) it is a single layer (no hidden layer), this topology provides an ability of simple implementation
and fast learning speed compared to other networks with hidden layers; and iv) the propagation of the signals
in the network will be in one direction only (feedforward), therefore each neuron will depend on the directed
input signals only. There is no activation function as in the conventional Hebbian neural network to
overcome the problem of selecting which one of the activation functions that gives the better results in
learning process.
The modified neural network has X vector (X: X1, X2, …, Xm) as an input of plaintext in ASCII
form, while the vector (Y:Y1, Y2, …, Ym) is the output of encryption process (ciphertext). The weight (Wij)
plays two roles: the first one, it acts as a ciphered key of length (m2
), such that, if the message length is
15 characters, then there are 15*15=225 different weights (key length). The second role, it represents the
strength of connection between the (ith
) input node and the (jth
) output node also represent. Figure 3 shows the
MHNN structure.
Learner Neural
Initial message
Initial Wet.
Ciphering Neural Net
Plaintext Ciphering Keys
Ascii Code
Ciphering Text
DNA Compression

2759
Figure 3. Architecture of MHNN
3. RESEARCH METHOD
The proposal system deals with two sites: sender and receiver, each site includes a main phase all
cooperate to achieve the proposal goals of security. These phases are: ciphering phase implemented at sender
site.to encrypt the plain text in somehow, so that the encryption text could not read to anyone except the
intended person. Another phase is the deciphering implemented at receiver site, its role is an inverse of
ciphering phase
3.1. Sender site
At this site, the sender will prepare several stages for the ciphering phase. Each stage has its own
function, the stages cooperated between them to result the encryption text before transferred to the receiver
site. The stages of ciphering phase are: Pre-processing stage, Ciphering stage, and DNA compression stage as
shown in Figure 4.
Figure 4. Ciphering phase
3.1.1. Pre-processing stage
This stage includes the following steps: i) prepare an initial (weight, message) and ii) assign each
character in plaintext message and initial message to its ASCII code.
3.1.2. Ciphering stage
To cipher the plaintext message using MHNN; two types of Hebbian nets will be used, these nets
are the Learner net and Ciphering net. For each net type has a specific function. The learner net function is to
prepare an optimal weights to the ciphered net as an input to this net, the following steps will conduct:
a. Learner neural net (LNN)
During the learning process, the following steps will be prepared:
− Input initial message as an input for LNN. The message must have the same length of the original
plaintext message.
− Normalize the initial weights to avoid the problem of overflow in the weight values. The normalization
process is applied to all initial weights to find optimal weights (ciphered keys), the normalization process
implemented by the following formula:
X3
X1
X2
Xn
Y1
Y2
Y3
Yn
Input Output
nn
Pre-processing
Stage
DNA compression
Stage
Ciphering
Stage

 ISSN: 2088-8708
2760
NWIJ=
Wij
∑m
i=1 ∑ Wij
m
j=1
(1)
b. Ciphering neural net (CNN)
The CNN net is used as a ciphering model for coding the plaintext as an output following the steps:
− Convert the plaintext to ASCII code as one of the inputs to CNN
− The optimal weights (ciphered keys) resulted from LNN will be the initial weights to CNN
− The ciphered message (Yj) is found by using formula (2):
Yj=∑ (NWij ∗ Xi)
m
i=1
𝑗 = 1, 2, 3, 4, . . . , 𝑚 (2)
− The weight (ciphered keys) is calculated by formula (3):
Wij=*Yj*[Xi-∑ (NWij ∗ Yk)
i
k=1 ]
𝑖, 𝑗 == 1, 2, 3, 4, … . , 𝑚 (3)
where  is learning factor (positive and <1), Wij are the updated weight. Xi is plain message (ASCII
Code), and Yj is ciphering message (ASCII Code).
3.1.3. DNA compression stage
The DNA is the abbreviation of deoxyribonucleic acid, it represents the whole life forms. DNA is
made of nucleotides and is a type of macromolecule that is biological. There are four groups of bases
corresponding to four types of nucleotides, they are adenine (A) and thymine (T) or cytosine (C) and guanine
(G). This DNA characterized by massive storage, parallelism, and hard cracking. These traits encouraged
researchers to move towards incorporating DNA into scientific research with many different directions that
include most of the sciences discipline [10]–[26].
a. DNA binary coding
The four chemical bases that make up DNA sequence A, C, G, and T bases, where biologically A is
connected to T, and C is connected to G. These synthesis rules can be modified in binary arithmetic by
changing input such as assuming that T is related to C or T is related to G. Using a binary encoding rule to
translate a hidden message into DNA rules. For each rule such as (A), the corresponding binary formulas can
be 00, 01, 10, or 11. for each DNA base. The encoding of DNA and its random properties make it an ideal
candidate for both coding and encoding. As a result, converting DNA into the binary form will result in
4!=24 different encoding methods On DNA bases, logical operations such as addition, subtraction, XOR,
AND, OR, and NOT are possible [27]–[29]. To compress/decompress the resulted ciphered text from the
previous stage, a DNA computing operation will used such as DNA Addition and DNA Subtraction
operations using the following steps:
− Convert the ciphered message to DNA coding using Table 1.
− Compress the converted DNA coding message by apply DNA addition operation using Table 2.
− Decode the compressed DNA message
− Send the decoded message to the receiver
3.2. Receiver site
Retrieving the original plaintext message, the receiver has to decrypt the receiving ciphered message
using deciphering phase. The receiver will follow a reverse sender steps with some exceptions such as using
a subtraction DNA computing operation as in Table 3 instead of addition to decompress the received
message. The deciphering phase consist of the stages bellow in order to get the original plaintext, these stages
illustrated in Figure 5.
Table 1. DNA coding
Base Name Binary Coding
A 00
C 01
G 10
T 11
Table 2. DNA addition [30]
+ T A C G
T C G T A
A G C A T
C T A C G
G A T G C
Table 3. DNA Subtraction [30]
- T A C G
T C G T A
A A C G T
C T A C G
G G T A C

2761
Figure 5. Receiver site phases
3.2.1. DNA decompression stage
− Code the compressed message to DNA using Table 1.
− Decompress the converted DNA message by apply DNA subtraction operation using Table 3.
− Convert the DNA message to binary form using Table 1.
− Convert the binary message to ASCII form.
3.2.2. Decryption stage
− Neural network based decryption module is used to retrieve the original message (plain text).
− To carry out the decryption process, using (4), where the sender and receiver sharing the final optimal
weights (ciphered keys):
Xj=∑ ((𝐼𝑊)𝑖𝑗 ∗ 𝑌𝑖)
𝑚
𝑖=1
j=1, 2, 3, 4, ..., m (4)
where Yi are the input in ASCII form of the ith
element in ciphered message, IWij is the inverse weight
that represents the connection between the ith
element in ciphertext and jth
element in plaintext. Xj is the
output of jth
element in the plaintext.
Finally, convert each ASCII value to its represented character to retrieve the original message. To
illustrate the proposed system thought, the example bellow taken to explain the proposal work steps included
the main phases:
A text message "Text Encryption Using Modified Hebbian Neural Net" is to be encrypted using MHNN",
with =0.5, and initial text "This text is used as one of initial inputs", as an input to the learner net.
During the cipher phase: Convert the two messages to their ASCII code:
"Text Encryption Using Modify Hebbian Neural Net" is to be encrypted using MHN
84 101 120 116 32 69 110 99 114 121 112 116 105 11 110 32 85 115 105 110 103 32 77 111 100 105 102
121 32 72 101 98 98 97 105 110 32 78 101 117 114 97 108 32 78 101 115.
This text is used as a one of the initial inputs (initial message input to LNN):
84 104 105 115 32 116 101 120 116 32 105 115 32 117 115 101 100 32 97 115 32 97 32 111 110 101 32 111
102 32 116 104 101 32 105 110 105 116 105 97 108 32 105 110 112 117 116 115.
The result of ciphering is:
1145 1160 1142 1065 1118 1030 1076 1271 1117 1054 1250 1272 1166 1097 1212 1066 1097 1022 996
1230 1255 1202 1201 1170 1165 1141 1212 1181 1315 1020 1281 1201 1166 1256 1259 1134 1121 1060
1076 1197 1070 1155 1013 1143 1224 1051 1241.
Each number consist of 4 digits, these numbers will be separated in two parts, if the number consist of 3
digits it will completed to 4 digits by adding 0 in left side as a in number 996:
11 45, 11 60, 11 42, 10 65, 11 18, 10 30, 10 76, 12 71, 11 17, 10 54, 12 50, 12 72, 11 66, 10 97, 12 12, 10 66,
10 97, 10 22, 09 96, 12 30, 12 55, 12 02, 12 01, 11 70, 11 65, 11 41, 12 12, 11 81, 13 15, 10 20, 12 81, 12 01,
11 66, 12 56, 12 59, 11 34, 11 21, 10 60, 10 76, 11 97, 10 70, 11 55, 10 13, 11 43, 12 24, 10 51, 12 41.
Each part will be converted to a binary form of 8 bits, if one part is results less than 4 bits then embeds it with
0's on left side to be 8 bits:
00001011 00101100, 00001011 00111100, 00001011 00101010, 00001010 01000001, 00001011 00010010,
00001010 00011110, 00001010 01001100, 00001100 01000111, 00001011 00010001, 00001010 00110110,
00001100 00110010, 00001100 01001000, 00001011 01000010, 00001010 01100001, 00001100 00001100,
00001010 01000010, 00001010 01100001, 00001010 00010110, 00001001 01100000, 00001100 00011110,
00001100 00110111, 00001100 00000010, 00001100 00000001, 00001011 01000110, 00001011 01000001,
00001011 00101001, 00001100 00001100, 00001011 01010001, 00001101 00001111, 00001010 00010100,
00001100 01010001, 00001100 00000001, 00001011 01000010, 00001100 00111000, 00001100 00111011,
00001011 00100010, 00001011 00010101, 00001010 00111100, 00001010 01001100, 00001011 01100001,
00001010 01000110, 00001011 00110111, 00001010 00001101, 00001011 00101011, 00001100 00011000,
00001010 00110011, 00001100 00101001.
The binary bits will be coded to DNA bases according to Table 1:
DNA decompression Stage Original plaintext stage
Decryption Stage

 ISSN: 2088-8708
2762
AAGT AGTA, AAGT ATTA, AAGT AGGG, AAGG CAAC, AAGT ACAG, AAGG ACTG, AAGG
CATA, AATA CACT, AAGT ACAC, AAGG ATCG, AATA ATAG, AATA CAGA, AAGT CAAG,
AAGG CGAC, AATA, AATA, AAGG CAAG, AAGG CGAC, AAGG ACCG, AAGC CGAA, AATA
ACTG, AATA ATCT, AATA AAAG, AATA AAAC, AAGT CACG, AAGT CAAC, AAGT AGGC,
AATA, AATA, AAGT CCAC, AATC AATT, AACC ACCA, AATA CCAC, AATA AAAC, AAGT
CAAG, AATA ATGA, AATA ATGT, AAGT AGAG, AACCT ACCC, AAGG ATTA, AAGG CATA,
AAGT CGAC, AAGG CACG, AAGT ATCT, AAGG AATC, AAGT AGGT, AATA ACGA, AAGG
ATAT, AATA AGGC.
The compress the message, adding DNA Computing operations are used, using Table 2:
CTTG, CGAG, CTCA, ACTG, CATA, CAAC, ACAT, ACTG, CATT, CGGC, CGGT, ACAC, ACTA,
ATTG, CCCC, ACTC, ATTG, CAGC, ATTC, CACC, CGTG, CCGT, CCGA, ACGA, ACTT, CTCT,
CCCC, AATT, CCCT, CACA, AAGA, CCGA, ACTA, CGAC, CGAG, CTTA, CACT, CGAT, ACAT,
ATTT, ACGC, CGGC, CCAG, CTCC, CAAC, CGTA, CTAA.
Convert the binary bits to decimal and transfer the decimal numbers the receiver:
126, 98, 130, 30, 72, 66, 19, 30, 79, 105, 106, 17, 44, 62, 85, 29, 62, 73, 69, 110, 91, 88, 24, 31, 112, 85, 15,
87, 68, 8, 88, 28, 97, 98, 124, 71, 99, 19, 63, 25, 105, 82, 117, 65, 108, 112.
4. RESULTS AND DISCUSSION
This paper introduced a modified cipher system that use the concepts of neural network and DNA
Computing to provide robust security. In this work, modernization of the HNN with no activation function so
the secret message will be encrypted by unsupervised neural network method. The results in Table 4 and
Figure 6 demonstrate the encryption/decryption processes are performed with a reasonable amount of time in
recovering the original message and achieved the goal of cannot break the transmitted ciphertext by the
intruder because the ciphertext is ciphered using a very long key.
Table 4. Encryption and decryption time
Characters/Message Encryption Time (ms) Decryption Time (ms)
47
400
1000
8445
125634
0.000011417
0.000131223
6.113423341
10.51000012
13.50230144
0.00001100
0.0002112016
0.117343320
0.151867112
0.198740113
Figure 6. Encryption and decryption time
4.1. DNA reference sequence
There are 163 million DNA reference sequences. The probability of the attacker making a guess is
(1/24). The sender not restricted in choosing any binary coding form for nucleotide bases. Hence, A can be
(00, 01, 10, 11) or T can be (00, 01, 10, 11) and so on for reset bases. Therefore, the number of binary coding
rules are 4×3×2×1=24. So, the chance of the attacker making the right guess is (1/24). As a consequence, the
likelihood of an attacker making a correct and accurate guess is ((1/24)*(1/24)*(1/(163*106
))).
0
5
10
15
47 400 1000 8445 125634
Encryption/Decryption Time
Encryption Time (ms) Decryption Time (ms)
Time
No. of Characters

2763
4.1.1. Randomness
A random process is one whose results are not understood. Intuitively, this is why randomness is
vital to our work because it offers a way to generate knowledge that an enemy cannot learn or predict. In our
work, we get different keys for each iteration. As a consequence, the attacker cannot predict the key.
4.1.2. Data compression
This section explain the result of using DNA compression rate for the amount of time it takes either
for encryption or decryption according to the numbers of characters that constitute the message (plain text) as
shown in Table 5 compared with research [31].
Table 5. Compression between [31] and the proposal using standard compression algorithms
Data Set LZFG HUFF RLE SF AC Proposal
Bib 2.90 8.16 5.26 5.56 5.23 2.33
Book1 3.62 8.17 4.57 4.83 4.55 2.55
Book2 3.05 8.17 4.83 5.08 4.78 2.66
News 3.44 7.98 5.24 5.41 5.19 2.54
Paper1 3.03 8.12 5.09 5.34 4.98 2.51
Paper2 3.16 8.14 4.68 4.94 4.63 2.65
progc 2.89 8.10 5.33 5.47 5.23 2.20
5. CONCLUSION
It is a challenge to maintain big data of an enormous population and protect this data. In this paper,
adopted new encryption method using hybrid techniques such as machine learning and DNA computing
operations have an important role in implementing secure environment. Using two modified HNN as a
ciphering model. The ciphering process based on the plain message and the number of epochs that based on
the used parameters. Hence, determine the actual parameters like initial weight, learner function for
encryption and decryption methods is hard for the intruder to guess. On other hand, visibility is the main
attribute of the DNA sequences, so finding the hidden message from a DNA sequence is difficult. Although,
using DNA computing operations for data compression increasing the confusion image and decrease the
transferring time compared with the algorithms shown in Table 5. In the future, one can use the proposed
model to transmit multimedia data such as images, audio and videos.
REFERENCES
[1] O. G. Abood and S. K. Guirguis, “DNA computing and its application to information and data security field: a survey,”
International Journal of Academic Engineering Research (IJAER), vol. 3, no. 1, pp. 1–5, 2019.
[2] A. Gehani, T. LaBean, and J. Reif, “DNA-based cryptography,” in DNA Based Computers V, American Mathematical Society,
2000, pp. 233–249.
[3] M. Barakat, C. Eder, and T. Hanke, “An introduction to cryptography,” mathematik, 2018. https://www.mathematik.uni-
kl.de/~ederc/download/Cryptography.pdf (accessed Sep. 20, 2018).
[4] A. Yayik and Y. Kutlu, “Neural network based cryptography,” Neural Network World, vol. 24, no. 2, pp. 177–192, 2014, doi:
10.14311/NNW.2014.24.011.
[5] “Hebb, D. O. The organization of behavior: A neuropsychological theory. New York: John Wiley and Sons, Inc., 1949. 335 p.
$4.00,” Science Education, vol. 34, no. 5, pp. 336–337, Dec. 1950, doi: 10.1002/sce.37303405110.
[6] P. Mishra, “What is Hebbian Learning?,” Data Driven Investor. 2019.
[7] G. Amato, F. Carrara, F. Falchi, C. Gennaro, and G. Lagani, “Hebbian learning meets deep convolutional neural networks,” in
Lecture Notes in Computer Science, Springer International Publishing, 2019, pp. 324–334.
[8] G. Lagani, G. Amato, F. Falchi, and C. Gennaro, “Training convolutional neural networks with Hebbian principal component
analysis,” Computer Vision and Pattern Recognition, Dec. 2020.
[9] A. Magotra and J. Kim, “Improvement of Heterogeneous transfer learning efficiency by using Hebbian learning principle,”
Applied Sciences, vol. 10, no. 16, p. 5631, Aug. 2020, doi: 10.3390/app10165631.
[10] S. Gangadharan and K. Raman, “The art of molecular computing: whence and whither,” Biomolecules, Feb. 2021.
[11] M. N. Sadiku, A. E. Shadare, and S. M. Musa, “DNA computing made simple,” Journal of Scientific and Engineering Research,
vol. 3, no. 2, pp. 116–118, 2016.
[12] S. G. Abels and E. F. Khisamutdinov, “Nucleic acid computing and its potential to transform silicon-based technology,” DNA and
RNA Nanotechnology, vol. 2, no. 1, Jan. 2015, doi: 10.1515/rnan-2015-0003.
[13] S. S. Kulkarni, “Implementation of SDES and DES using neural network,” International Journal of Innovations in Engineering
and Technology (IJIET), vol. 1, no. 2, pp. 34–49, 2012.
[14] S. D. Jagtap, P. BalaRamudu, and M. K. Singh, “Cryptography based on artificial neural network,” International Journal of
Advanced Research in Electronics and Communication Engineering (IJARECE), vol. 4, no. 11, pp. 2785–2789, 2015.
[15] G. S. Mohammed, “Text encryption algorithm based on chaotic neural network and random key generator,” Text Encryption
Algorithm Based on Chaotic Neural Network and Random Key Generator, 2016.
[16] S. S. Roy, S. A. Shahriyar, M. Asaf-Uddowla, K. M. R. Alam, and Y. Morimoto, “A novel encryption model for text messages
using delayed chaotic neural network and DNA cryptography,” in 2017 20th International Conference of Computer and
Information Technology (ICCIT), Dec. 2017, pp. 1–6, doi: 10.1109/ICCITECHN.2017.8281796.
[17] P. Dixit, M. C. Trivedi, A. K. Gupta, V. K. Yadav, Vineet, and K. Singh, “Video steganography using concept of DNA sequence

 ISSN: 2088-8708
2764
and index compression technique,” International Journal of Engineering and Advanced Technology (IJEAT), vol. 8, no. 5,
pp. 408–417, 2019.
[18] S. Basu, M. Karuppiah, M. Nasipuri, A. K. Halder, and N. Radhakrishnan, “Bio-inspired cryptosystem with DNA cryptography
and neural networks,” Journal of Systems Architecture, vol. 94, pp. 24–31, Mar. 2019, doi: 10.1016/j.sysarc.2019.02.005.
[19] S. Namasudra, D. Devi, S. Kadry, R. Sundarasekar, and A. Shanthini, “Towards DNA based data security in the cloud computing
environment,” Computer Communications, vol. 151, pp. 539–547, Feb. 2020, doi: 10.1016/j.comcom.2019.12.041.
[20] H. Li et al., “The specific DNA barcodes based on chloroplast genes for species identification of Orchidaceae plants,” Scientific
Reports, vol. 11, no. 1, p. 1424, Dec. 2021, doi: 10.1038/s41598-021-81087-w.
[21] S. P. Singh and M. E. Naidu, “A Novel method to secure data using DNA sequence and Armstrong Number,” Asian Journal For
Convergence In Technology (AJCT), vol. 3, no. 3, 2017.
[22] P. Malathi, M. Manoaj, R. Manoj, R. Vaikunth, and R. E. Vinodhini, “Highly improved DNA based steganography,” Procedia
Computer Science, vol. 115, pp. 651–659, 2017, doi: 10.1016/j.procs.2017.09.151.
[23] M. Indrasena Reddy, A. P. Siva Kumar, and K. Subba Reddy, “A secured cryptographic system based on DNA and a hybrid key
generation approach,” Biosystems, vol. 197, Nov. 2020, doi: 10.1016/j.biosystems.2020.104207.
[24] E. Volna, M. Kotyrba, V. Kocian, and M. Janosek, “Cryptography based on neural network,” in ECMS 2012 Proceedings edited
by: K. G. Troitzsch, M. Moehring, U. Lotzmann, May 2012, pp. 386–391, doi: 10.7148/2012-0386-0391.
[25] H. Al-Mahdi, M. Alruily, O. R.Shahin, and K. Alkhaldi, “Design and analysis of DNA encryption and decryption technique based
on asymmetric cryptography system,” International Journal of Advanced Computer Science and Applications, vol. 10, no. 2,
2019, doi: 10.14569/IJACSA.2019.0100264.
[26] M. Gao, A. Krissanaprasit, A. Miles, L. C. Hsiao, and T. H. LaBean, “Mechanical and electrical properties of DNA hydrogel-
based composites containing self-assembled three-dimensional nanocircuits,” Applied Sciences, vol. 11, no. 5, Mar. 2021, doi:
10.3390/app11052245.
[27] B. R, D. S, and P. K, “Secure data transmission using DNA sequencing,” IOSR Journal of Computer Engineering, vol. 16, no. 2,
pp. 19–22, 2014, doi: 10.9790/0661-16221922.
[28] E. I.AbdEl-Latif and M. I. Moussa, “Chaotic information-hiding algorithm based on DNA,” International Journal of Computer
Applications, vol. 122, no. 10, pp. 38–42, Jul. 2015, doi: 10.5120/21740-4949.
[29] Adithya B. and Santhi G., “DNA computing using cryptographic and steganographic strategies,” in Data Integrity and Quality,
IntechOpen, 2021.
[30] Z. Ahmad, H. G. Umar, C. Li, and L. Chen, “A DNA-based security solution using aggregated chaos cross and cubic map,”
International Arab Journal of Information Technology, vol. 13, pp. 873–879, 2016.
[31] K. Sailunaz, M. R. A. Kotwal, and M. N. Huda, “Data compression considering text files,” International Journal of Computer
Applications, vol. 90, no. 11, pp. 27–32, Mar. 2014, doi: 10.5120/15765-4456.
BIOGRAPHIES OF AUTHORS
Sahar Adill Kadum received the Ph.D. degrees in computer science from higher
institute of informatics and computer science. Currently, she is an Associate Professor at the
Department of computer science at Babylon University–collage science for women. Her
research interests include information security, bioinformatics security. Image security. She
can be contacted at email: dr.sahar.adill@gmail.com, wsci.sahar.adil@uobabylon.edu.iq.
Ali Yakoob Al-Sultan received a PhD in Artificial Intelligent from University of
Babylon Currently, he is an lecture at the Department of computer science at Babylon
University–collage science for women. His research interests include Artificial Intelligent
applications, machine learning, and deep learning. He can be contacted at email:
wsci.ali.yakoob@uobabylon.edu.iq.
Najlaa Adnan Hadie received a BSc and high diploma in mathematics and
operation research from University of Babylon Currently, she is an MSc student at the
Department of Mathematic science at Babylon University. Her research interests in operation
research and computation research. She can be contacted at email: hudinajlaa96@gmail.com.

Data protection based neural cryptography and deoxyribonucleic acid

More Related Content

Similar to Data protection based neural cryptography and deoxyribonucleic acid (20)

More from IJECEIAES (20)

Recently uploaded (20)

Data protection based neural cryptography and deoxyribonucleic acid