Development of an Electronic Voting System using Blockchain
Technology and Deep Hybrid Learning
MD. ABDUL BASED1,*, ELIAS UR RAHMAN1, MOHAMMAD SHORIF UDDIN2
1Department of Computer Science and Engineering,
Dhaka International University,
Dhaka,
BANGLADESH
2Department of Computer Science and Engineering,
Jahangirnagar University,
Dhaka,
BANGLADESH
*Corresponding Author
Abstract: - Democratic people cannot function properly in today's sophisticated societies (where voting is a
prominent issue) without electronic voting technologies. This study explores the use of hybrid learning
algorithms for biometric authentication of voters, and blockchain technology for secure electronic voting. The
thorough analysis includes a collection of more than 50,000 fingerprint samples using custom Convolutional
Neural Network (CNN), VGG16, VGG19, Xception, and Inception. The algorithms are evaluated using F1-
score, recall, accuracy, and precision. By combining Random Forest with a specially designed CNN, a novel
hybrid learning algorithm is developed for authentication purposes. This blended model provides the best
outcome in terms of accuracy (99.32%) and precision (99.32%). In addition, a web application was developed.
This application integrates blockchain technology for electronic voting using Flask, HTML, and Solidity. By
using blockchain, tampering and unauthorized access are prevented. It also ensures impartial voting and secure
storage. The tabular presentation of the results provides a clear summary of each candidate's total votes.
Key-Words: - Electronic Voting, Hybrid Learning Algorithms, Biometric Fingerprint Matching, Voting
Security, CNN, Random Forest.
Received: November 16, 2023. Revised: June 19, 2024. Accepted: July 14, 2024. Published: August 14, 2024.
1 Introduction
Voting is one of the many aspects that are rapidly
changing due to new technologies. Making sure that
the votes are secure, reliable, and visible to all is
essential as paper ballots are transited to electronic
ones in Electronic Voting (e-voting). The goal of
this paper is to integrate two significant
technological advancements. One method is
matching fingerprints using hybrid learning
algorithms. The other is improving and securing
electronic voting with blockchain technology. While
e-voting offers great potential, there are drawbacks
as well. The key concerns here are voters’ eligibility
and ensuring that the people casting votes are the
actual voters. Using biometric identification
techniques appears promising in this scenario.
Significant studies [1], [2] have sparked interest in
the safe and accurate use of fingerprints for voter
identity confirmation. Like most other fields, voting
electronically with fingerprints resolves the unique
issues associated with selecting public officials.
As biometric identification becomes popular in
e-voting, picking the right techniques is very
important. Many researchers have seen the use of
hybrid learning methods. These mixed ways can
provide better results. The biometric system
analyzed in this paper includes VGG16, VGG19,
Xception, Inception, and a modified Convolutional
Neural Network (CNN). Each of these algorithms is
applied to spot special shapes in fingerprint
information. This makes the task of matching
fingerprints better and quicker. The reason for using
mixed learning methods is because the blended
method can pick out important details from hard
data, helping to see unique features in each
fingerprint. This research wants to use the good
points of hybrid learning in different ways, [3].
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
194
Volume 23, 2024
Blockchain technology has become a big talk
for its chance to change e-voting systems and make
them safer and more open. Blockchain allows
people to keep records without a big boss or any
person changing the information, and offers a new
way to deal with old problems that come up in
traditional voting methods. An explanation of how
blockchain can change things are discussed in [4]
and [5]. It doesn't just keep voting data safe but also
creates a permanent record of the whole election
process. The blockchain to the voting system is
implemented in this paper by creating a website
using Flask, HTML, and Solidity. This application
connects the voting system on computers with
blockchain. It makes sure votes are safe, can be
checked easily, and counted clearly. This study
wants to fix the worries about vote rigging, sneaky
access, and dishonest actions in elections by using
blockchain's security features.
There are two main goals of this research. First,
it plans to carefully study the chosen mixed learning
methods (VGG16, VGG19, Xception, Inception,
and CNN) to see how well they work for matching
fingerprints. This approach looks to measure quality
including accuracy, precision, recall, and F1-score.
Second, the study tries to show how blockchain
technology can be added to the design of the voting
system. This gives a safe and open way for storing
and counting votes. Working on the website
application aims to make online voting systems
stronger and grow trust in democracy.
2 Literature Review
The path of Electronic Voting (e-voting) systems
has changed over many years. It grew together with
progress in technology and the change society made
towards becoming more digital. In the middle of the
1900s, punch cards were first used in e-voting
systems. These are like beginning steps for having
e-voting everywhere instead of paper ballots in
polling stations.
In the late 1900s, fingerprints became a main
way to find criminals after it was found out that the
fingerprints were unique, [1]. This important and
useful finding soon took over old ways, like using
body size to find out who did bad things. Using
fingerprints for checking bad people got popular.
The police in many cases used them and made a
database with fingerprints from the criminals. Then
they started using lifted fingerprints found at crime
scenes to know who did the crime and catch them.
Biometric recognition has changed a lot in the past
century. It's now used for safe logins and checks,
making sure people have certain jobs. It's also used
by legal authorities to figure out who was involved
in a situation or check positive ID of individuals in
many areas.
A combination of Convolutional Neural
Network (CNN) and Back Propagation Neural
Network (BPNN) is proposed in [6] for an improved
Fingerprint Identification System. Although BPNN
is a wonderful choice for hybrid learning, some of
its drawbacks take it to the edge. As a result, CNN
is combined with Random Forest (RF) in this paper
to create a more suitable and robust hybrid learning
algorithm. This gives freedom and is more suitable
for hierarchical structures.
In [7], a method is proposed where the authors
combined a Support Vector Machine (SVM) with
CNN. Using their proposed method, they acquired
an accuracy of 95%. However, Random Forest is
more suited for matching fingerprint algorithms. As
a result, in this study, using the combination of RF
and CNN gives an accuracy of more than 99%.
In [8], another method is presented where for
image preprocessing Improved Whale Optimization
Algorithm (IWO), and for person detection, Teacher
Learning-based Deep Neural Network (TL-DNN)
are used. This proposed method gives the study a
new edge as it combines different algorithms for the
purpose of clear biometric identification.
Hybrid learning, a fusion of different machine
learning architectures, has gained prominence in
various applications, including biometric
identification. A seminal overview is provided in [3]
of deep learning, highlighting the capabilities of
Neural Networks (NN) in extracting intricate
features from complex datasets. In the context of
electronic voting, hybrid learning algorithms offer
the promise of improved accuracy and robustness in
biometric matching.
Decentralized and tamper-resistant ledger
blockchain has emerged as a revolutionary
technology with remote references for the security
of e-voting systems. The work in [4] clarifies the
basic principles of blockchain and its potential
applications in various areas, including elections.
Blockchain technology can handle vote tampering,
fraud, and unauthorized access in the voting
systems.
Blockchain mining along with the mechanisms
to offer the security of blockchain networks are
discussed in [5]. The integration of blockchain in e-
voting ensures the security of the voter and the
verifiability of the vote by the voter. The blockchain
also mitigates the vulnerabilities associated with
centralized voting databases and provides a
decentralized trusted system for recording and
counting votes.
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
195
Volume 23, 2024
The hybrid learning algorithms for fingerprint
matching in this research provide better
performance compared to the relevant existing
works. The evaluation of the VGG16, VGG19,
Xception, Inception, and CNN algorithms is done
using the performance metrics accuracy, precision,
recall, and F1-score. This provides viability in real-
world e-voting scenarios.
The blockchain technology in e-voting adds an
additional level of security and transparency through
its decentralized and tamper-resistant nature. This
ensures that the votes are stored securely, verifiable,
and tallied transparently in order to address the
integrity of the e-voting system. Thus this research
enhances the security, reliability, and transparency
of e-voting systems by combining both biometric
identification and blockchain technology.
3 Proposed Methodology
This study creates a new path to the future using
efficient algorithms to identify fingerprint matching.
Voter identification and voting security are some of
the greatest threats to democratic countries. This
paper tries to solve them using various methods
containing hybrid deep learning algorithms for voter
identification using fingerprints, and blockchain
technology for the security of the voting system.
Thus, the path to a secure and trustworthy voting
system is clearer. The hybrid deep learning model
for voter authentication is shown in Figure 1.
The algorithm for the hybrid learning model is
shown in Algorithm 1.
Algorithm1: Algorithm for the hybrid deep learning
algorithm
1. Loading the dataset into a single variable as
a multi-dimensional array
2. The arrays are shuffled randomly to train
the model efficiently
3. Separating the training image and training
labels
4. Converting the arrays for training
5. Creating or importing the model
6. Training the model
7. Extracting the features from the images
8. Using the features to train the Random
Forest Classifier
9. Evaluating the hybrid model
3.1 Data Collection
Data Collection is a crucial part of the study as it
vastly depends on the collected image data for
evaluation of the algorithms. For this purpose, a
team dataset of around 700 students was used. The
open dataset of SOCOFing [9] was also used for the
study. In the SOCOFing’s dataset, it had Altered
Fingerprint images. These Altered Fingerprint
images contained Arch, Left Loop, Right Loop,
Tented, and Whorl fingerprints. Combining this
dataset with the team dataset, the study worked with
upmost 50,000 fingerprint image datasets. All of
these fingerprint images were combined for a more
complex and versatile study. Figure 2 shows some
samples of the fingerprint images.
Fig. 1: The Hybrid Deep Learning Model for Voters’ Authentication
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
196
Volume 23, 2024
Fig. 2: Fingerprint Image Samples for Dataset
3.2 Data Preprocessing
The dataset is loaded as a multi-dimensional array.
Representing images as multi-dimensional arrays
allows thepreservation of the structure of the images
and makes it easier to interpret and process. Deep
learning frameworks like TensorFlow and Keras
work with multi-dimensional arrays as their primary
data structure. Loading image data as multi-
dimensional arrays makes it compatible with these
frameworks. Neural Network (NN) model,
including CNN, VGG16, VGG19, Xception, and
Inception, expect input data in the form of multi-
dimensional arrays. Each of these dimensions
corresponds to a specific aspect of the data, such as
the height, width, and channels for images.
Converting images to multi-dimensional arrays
allows them to normalize according to the needs of
the deep learning algorithms.
3.3 Neural Network Models for Feature
Extraction
Various NN models like CNN, VGG16, VGG19,
Xception, and Inception were used to extract the
features from the images so that the Random Forest
Classifier could be trained upon the features. CNN
is one kind of feed-forward neural network that is
able to extract features from data. CNN does not
extract features manually from the data as opposed
to traditional feature extraction methods, [10], [11],
[12]. This algorithm extracts the features using a
convolutional structure which is very efficient. It
can detect patterns in the images using these layers.
Convolutional layers apply filters to the input to
detect various features and pool layers down sample
the spatial dimensions.
VGG16 and VGG19 are models that have a
straightforward architecture with only 3x3
convolutional filters and 2x2 pooling layers, [13].
Although they might have a similar structure,
VGG19 is deeper than VGG16. However, this
increased depth comes at the cost of higher
computational requirements.
Xception short for “Extreme Inception” replaces
traditional convolutions with depth-wise separable
convolutions. Although this reduces the number of
parameters and computational cost it is quite
inefficient in the context of finding matches in the
fingerprint images.
Inception, also known as GoogLeNet, is a CNN
architecture that consists of multiple stacked
Inception modules, [13]. These modules allow the
network to capture features at different scales and
resolutions, enhancing its ability to recognize
complex patterns.
3.4 Random Forest
Random Forest (RF) is used to train the model after
extracting the features from the fingerprint image
data. Combining the RF with other Neural Networks
such as CNN, VGG16, VGG19, Xception, and
Inception creates a unique blend of robustness,
interpretability, and predictive accuracy, making it
suitable for enhancing the performance of hybrid
learning systems.
RF is a learning method based on decision tree
classifiers, [14]. As a result, it brings several key
points to the table which make it especially well-
suited for integration with NN. Particularly RF is
great in bringing the predictive power of multiple
decision trees together, creating a stack that is often
more resilient and accurate than individual models.
RF can also capture complex non-linear
relationships within the data, making them more
effective in the context where the relationships are
not easily modeled by linear algorithms. As a result,
the combination of RF and other NNs is used in this
paper to find out the best combination.
3.5 Evaluation Metrics
Several evaluation metrics such as accuracy,
precision, recall, and F1-score were used to test all
the applied algorithms in this paper. As a result, the
combined models were evaluated thoroughly.
3.5.1 Accuracy
Accuracy is one of the fundamental metrics for
evaluating the performance of machine learning
models by measuring the overall performance of the
predictions, [15]. Equation 1 represents accuracy as
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
197
Volume 23, 2024
the total percentage of correctly identified
fingerprints out of the total dataset.
 
 󰇛󰇜
3.5.2 Precision
Precision is one of the main evaluation metrics in
the biometric identification system deploying hybrid
learning fingerprint matching algorithms. The
accuracy of positively identified fingerprint
matching among all instances is measured by
precision, [15]. The precision that shows the model's
ability to correctly identify and classify the genuine
matches as True Positives (TP) while minimizing
False Positives (FP) is shown in Equation 2.
 
  󰇛󰇜
3.5.3 Recall
The ability of a model in order to correctly identify
and capture all relevant instances of positive cases
within the dataset is measured by Recall or
Sensitivity [15]. The Recall value is measured using
Equation 3 where a high value indicates that the
model performs better at identifying genuine
matches. It also minimizes instances of False
Negatives (FN) in which cases the original matches
are overlooked.
 
 󰇛󰇜
3.5.4 F1-score
A model’s performance can be assessed [15] using
F1-score which is measured using Equation 4. The
F1-score provides measurements of both false
positives and false negatives. A high value indicates
a desirable balance between the precision and recall
values.
   
  󰇛󰇜
Table 1. Performance metrics of the hybrid learning algorithms
Algorithm
Accuracy
Precision
F1 Score
VGG16
89.36
88.39
84.37
Vgg19
90.3
88.51
84.27
Xception
84.67
82.74
74.64
Inception
82.9
81.40
72.06
CNN
99.32
99.32
99.32
Fig. 3: Proposed E-voting System using Blockchain Technology
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
198
Volume 23, 2024
3.6 Blockchain-based E-voting System
Blockchain is a decentralized ledger that is most
common in crypto-currency. Basically in a
blockchain, the transactions are securely stored as a
chain of blocks.
This chain continuously grows as new blocks
are added after each transaction, [16]. Blockchain is
a revolutionary technology that highlights the need
for decentralized, persistent, anonymous, and
authenticated data security.
The technology integrates the core elements
such as cryptographic hash, digital signatures based
on asymmetric cryptography, and distributed
consensus mechanisms, [16]. It implies that
transactions without the need for a central authority
reduce costs and improve efficiency. As a result,
blockchain technology can be used in various
sensitive fields where data security is crucial. It can
be used in Financial Services, Healthcare Services,
Security Services, and most importantly Voting
Services. In this paper, the blockchain technology
was implemented using Flask, HTML, and Solidity.
Figure 3 shows the voting system using blockchain
technology. The voters are assumed to be
authenticated by fingerprint using the proposed
hybrid learning algorithm, and then voting is
conducted with the blockchain technology as shown
in Figure 3.
The use of blockchain technology ensures
temper-resistant voting records, which can establish
voters’ trust as the records are transparent and
immutable with the decentralized and cryptographic
features of the technology. The use of hybrid deep
learning algorithms enhances the security of the
framework by being able to detect anomalies and
fraudulent activities in real-life implementation. The
paper's innovative approach not only addresses the
core challenges of e-voting, but also opens up
avenues for further exploration in optimizing system
performance, scalability, and usability.
4 Results
This section shows the results of the experiments
along with a comparative analysis of different
algorithms that were applied.
4.1 Performance Evaluation of Hybrid
Learning Algorithms
The hybrid learning algorithms such as VGG16,
VGG19, Xception, Inception and CNN combined
with Random Forest was subjected to rigorous
evaluation using a dataset of over 50,000 fingerprint
samples. The evaluation metrics shown in Table 1,
including accuracy, precision, recall, and F1-score
provide insight into the effectiveness of each
algorithm in biometric fingerprint matching.
4.2 Comparative Analysis
Figure 4 shows the results of different hybrid
learning algorithms. Compared to other algorithms,
the combination of CNN and RF hybrid learning
algorithms provides the highest accuracy of 99.32%
for around 50,000 fingerprint images. This hybrid
algorithm also provides higher precision, recall, and
F1 scores. The combined CNN and RF had the
ability to discern intricate patterns in fingerprint
data, showing its superior performance and
efficiency in biometric identification. Overall the
other algorithms, the CNN and RF got the upper
hand because of the hierarchical feature learning of
the convolutional layers. CNN can find patterns
regardless of their positions in the image since it can
demonstrate translation invariance.
Fig. 4: Comparative Analysis of Hybrid Learning
Algorithms
Compared to Xception and Inception, VGG16
and VGG19 show better performance, indicating the
reliability of these architectures in fingerprint
matching. Though the performance of Xception and
Inception is slightly lower, these algorithms
demonstrate competitive results and have potential
areas for optimization and alternative applications.
Fig. 5: The Page after Casting the Vote by a Voter
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
199
Volume 23, 2024
Fig. 6: Adding the vote into the Blockchain
Fig. 7: Showing Voting Results
Table 2. Comparative Study of the Proposed Algorithm with other Published Works
4.3 Blockchain Integration
The integration of blockchain technology into the
electronic voting model demonstrates significant
advancements in security and transparency. The
decentralized and tamper-resistant nature of the
blockchain ensures that each vote is securely stored,
verifiable, and transparently tallied. The
cryptographic features of blockchain mitigate
several issues relevant to vote tampering,
unauthorized access, and overall electoral
misconduct. The results indicate that the
blockchain-based electronic voting model provides
an immutable and transparent ledger, addressing
long-standing challenges associated with centralized
voting databases.
Usability testing of the web application revealed
a user-friendly interface, allowing voters to securely
cast their votes with ease. The integration of
blockchain does not compromise the usability of the
electronic voting system, providing a seamless
experience for voters while enhancing the security
of the overall process. Figure 5, Figure 6 and Figure
7 demonstrate some snapshots of the blockchain
integration into the proposed e-voting system.
5 Discussion
This section compares the findings of the proposed
work with the existing related research works and
assesses the contributions. It also delves into the
practical use of the blockchain and explores
efficient consensus mechanisms.
References
Fingerprint classification technique
Accuracy (%)
[17]
Random Forest (RF) + CNN
96.75
[18]
ROIFE_CNN on Gabor filtering image of ROI
95.10
[19]
CNN
96.01
[20]
Decision trees
98.00
[21]
SVM algorithm + naive Bayes method
95.60
[22]
Random Forest algorithm
96.50
[23]
Support Vector Machine (SVM)
94.97
[24]
Fuzzy-neural network classifier
98.00
[25]
Fusion of CNN and Gabor features
99.87
Proposed Method
CNN (Feature Extraction) + RF
99.32
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
200
Volume 23, 2024
5.1 Comparative Study of the Proposed
Algorithms with other Published Works
The results of the hybrid learning algorithms for
biometric fingerprint matching were impressive. As
shown in Table 2, the CNN+RF algorithm
demonstrated exceptional accuracy, precision,
recall, and F1 score. The comparison with other
research papers validates the efficacy of the chosen
algorithms.
The proposed method, combining CNN for
feature extraction and RF, achieves a notable
accuracy of 99.32%. This accuracy is competitive
and comparable to other state-of-the-art techniques
presented in Table 2. The fusion of deep learning-
based feature extraction with the ensemble learning
capabilities of RF contributes to the high accuracy
of the proposed method. The utilization of CNN for
capturing hierarchical representations in fingerprint
images followed by RF for robust classification
demonstrates the effectiveness of the hybrid
approach in achieving superior accuracy compared
to individual methods.
Referring to Table 2, it is notable that in all the
published works the dataset size was much smaller
than the dataset used in this paper. The exception is
in [24] which uses more samples, however the
accuracy is lower (98%) compared to the proposed
method (99.32%). The work in [25] achieved more
accuracy (99.87%) than the proposed method
presented in this paper. However, they have used
only 500 samples whereas the proposed method
includes 50,000 samples.
In this research, the accuracy is found for the
applied deep hybrid learning algorithms. The
algorithm with the highest accuracy can be used in
any e-voting system for biometric identification.
Thus, it will enhance the security of the e-voting
system instead of using the traditional algorithms.
Furthermore, the integration of blockchain
technology also improves the security of voting and
counting and provides transparency.
5.2 Blockchain Systems
The integration of blockchain technology into the
electronic voting model aligns with and builds upon
the transformative potential highlighted in related
research papers. Works in [4] and [5] underscore the
foundational principles of blockchain—
decentralization and immutability—and their
applicability to secure diverse domains. The
findings of the study echo the sentiments of [4],
emphasizing the decentralized and tamper-resistant
nature of blockchain as a robust solution for
securing e-voting systems. The results demonstrate
the successful implementation of blockchain in
providing a secure and transparent ledger for storing
and tallying votes, mitigating concerns raised in [5]
regarding traditional centralized voting databases.
This paper not only contributes to the theoretical
underpinnings of blockchain security but also
addresses practical concerns related to usability.
Usability testing of the web application revealed a
user-friendly interface, affirming that the integration
of blockchain does not compromise the accessibility
or ease of use for voters.
6 Conclusion
This paper contributes to the advancement of online
voting by identifying a robust biometrics technique
and illustrating how blockchain technology can
further enhance voting security. By incorporating
these tools, the study paves the way for future
advancements in electronic voting systems that seek
to increase public acceptance of democracy,
accuracy, and transparency.
This research is an innovative attempt to use
cutting-edge technologies to advance electronic
voting systems. The study aimed to strengthen the
security and transparency of electronic voting by
combining the transformative potential of
blockchain technology with hybrid learning
algorithms, such as CNN with Random Forest,
Xception, Inception, VGG16, and VGG19, for
biometric fingerprint matching. The best hybrid
learning algorithm was CNN+RF, which
demonstrated the highest levels of accuracy,
precision, recall, and F1 score. These results
established CNN+RF as a dependable method for
validating voters. Contributing substantially to the
field, the research addressed critical gaps by
providing a comparative analysis of hybrid learning
models for biometric identification and integrating
blockchain to enhance security and transparency.
The study highlights the necessity of ongoing
usability testing and recommends that future
research concentrate on diverse datasets and
iterative usability improvements while
acknowledging limitations such as dataset
specificity. This study represents a major
advancement in electronic voting systems and
provides a framework for more investigation and
application. When combined with blockchain
integration, the identified CNN+RF algorithm
generates a model that anticipates the changing
requirements of democratic processes. In the future,
electronic voting will strike a balance between
security, usability, and adaptability, fostering
confidence and trust in the political process,
according to the study.
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
201
Volume 23, 2024
References:
[1] Jain, A. K., Ross, A., & Prabhakar, S. (2004).
An introduction to biometric
recognition. IEEE Transactions on circuits
and systems for video technology, vol. 14(1),
pp. 4-20. DOI: 10.1109/TCSVT.2003.818349.
[2] Jain, A. K., & Kumar, A. (2012). Biometric
recognition: an overview. Second generation
biometrics: The ethical, legal and social
context, The International Library of Ethics,
Law and Technology, vol 11. Springer,
Dordrecht. Pp. 49-79. DOI: 10.1007/978-94-
007-3892-8_3.
[3] LeCun, Y., Bengio, Y., & Hinton, G. (2015).
Deep learning. Nature, vol. 521(7553), pp.
436-444. DOI: 10.1038/nature14539
[4] Ishaan Swan, M. (2015). Blockchain:
Blueprint for a new economy. 1st Edition,
O'Reilly Media, Inc. ISBN-10:
9781491920497 ASIN: 1491920491.
[5] Kiayias, A., Koutsoupias, E., Kyropoulou, M.,
& Tselekounis, Y. (2016). Blockchain mining
games. In Proceedings of the 2016 ACM
Conference on Economics and Computation,
Melbourne, Australia (pp. 365-382). DOI:
https://doi.org/10.1145/2940716.2940773.
[6] Kumar, R., & Patil, M. E. (2022). Improved
Fingerprint Identification System Using
Hybrid Deep Learning. Industrial
Engineering Journal, vol. 15(11), pp. 24-30.
[7] SaiTeja, C., & Seventline, J. B. (2023). A
hybrid learning framework for multi-modal
facial prediction and recognition using
improvised non-linear SVM classifier. AIP
Advances, vol. 13 (025316), pp. 1-8. DOI:
https://doi.org/10.1063/5.0136623.
[8] Jadhav, A.B, Deshmukh, N.K., & Humbe, V.
T. (2023). HDL-PI: hybrid Deep Learning
technique for person identification using
multimodal finger print, iris and face
biometric features. Multimedia Tools and
Applications. Vol. 82(19), 30039-30064. DOI:
10.1007/s11042-022-14241-9.
[9] Sokoto Coventry Fingerprint Dataset. Kaggle.
Date: 29/12/2023, [Online].
https://www.kaggle.com/datasets/ruizgara/soc
ofing (Accessed Date: July 10, 2024).
[10] Lindeberg, T (2012). Scale Invariant Feature
Transform. Chapter 7(5):104249, in book
Scholarpedia (2012): 10491. DOI:
10.4249/scholarpedia.10491.
[11] Dalal, N., & Triggs, B. (2005). Histograms of
oriented gradients for human detection. IEEE
computer society conference on computer
vision and pattern recognition (CVPR'05),
San Diego, CA, USA (Vol. 1, pp. 886-893).
IEEE. DOI: 10.13140/RG.2.2.23788.85122.
[12] Ahonen, T., Hadid, A., & Pietikainen, M.
(2006). Face description with local binary
patterns: Application to face
recognition. IEEE transactions on pattern
analysis and machine intelligence, vol.
28(12), 2037-2041. DOI:
10.1109/TPAMI.2006.244.
[13] Li, Z., Liu, F., Yang, W., Peng, S., & Zhou, J.
(2021). A survey of convolutional neural
networks: analysis, applications, and
prospects. IEEE Trans Neural Netw Learn
Syst. Vol. 33(12) pp. 6999-7019. DOI:
10.1109/TNNLS.2021.3084827.
[14] Rodriguez-Galiano, V. F., Ghimire, B.,
Rogan, J., Chica-Olmo, M., & Rigol-Sanchez,
J. P. (2012). An assessment of the
effectiveness of a random forest classifier for
land-cover classification. ISPRS journal of
photogrammetry and remote sensing, vol. 67,
pp. 93-104. DOI:
https://doi.org/10.1016/j.isprsjprs.2011.11.002
[15] Yacouby, R., & Axman, D. (2020).
Probabilistic Extension of Precision, Recall,
and F1 Score for More Thorough Evaluation
of Classification Models. Proceedings of the
First Workshop on Evaluation and
Comparison of NLP Systems, pp. 79-91.
DOI:10.18653/v1/2020.eval4nlp-1.9.
[16] Zheng, Z., Xie, S., Dai, H. N., Chen, X., &
Wang, H. (2018). Blockchain challenges and
opportunities: A survey. International journal
of web and grid services, vol. 14(4), 352-375.
DOI: 10.1504/IJWGS.2018.095647.
[17] Nguyen, H. T., & Nguyen, L. T. (2019).
Fingerprints classification through image
analysis and machine learning
method. Algorithms, vol. 12(11), 241, pp.1-
11. DOI: https://doi.org/10.3390/a12110241.
[18] Yang, J., Wu, Z., & Zhang, J. (2018). A
robust fingerprint identification method by
deep learning with Gabor filter
multidimensional feature expansion. In: Sun,
X., Pan, Z., Bertino, E. (eds) Cloud
Computing and Security. ICCCS 2018,
Nagoya, Japan. Lecture Notes in Computer
Science, vol. 11065. Springer, Cham. DOI:
https://doi.org/10.1007/978-3-030-00012-
7_41.
[19] Michelsanti, D., Ene, A. D., Guichi, Y., Stef,
R., Nasrollahi, K., & Moeslund, T. B. (2017).
Fast fingerprint classification with deep neural
networks. In International Conference on
Computer Vision Theory and Applications,
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
202
Volume 23, 2024
Porto, Portugal (pp. 202-209). SCITEPRESS
Digital Library. DOI:
10.5220/0006116502020209.
[20] Everingham, M., Van Gool, L., Williams, C.
K., Winn, J., & Zisserman, A. (2010). The
pascal visual object classes (voc)
challenge. International journal of computer
vision, vol. 88, pp. 303-338. DOI:
https://doi.org/10.1007/s11263-009-0275-4.
[21] Hong, J. H., Min, J. K., Cho, U. K., & Cho, S.
B. (2008). Fingerprint classification using
one-vs-all support vector machines
dynamically ordered with naı¨ve Bayes
classifiers. Pattern Recognition, vol. 41(2),
pp. 662-671. DOI:
10.1016/j.patcog.2007.07.004.
[22] Strobl, C., Malley, J., & Tutz, G. (2009). An
introduction to recursive partitioning:
rationale, application, and characteristics of
classification and regression trees, bagging,
and random forests. Psychological
methods, vol. 14(4), 323-348. DOI:
10.1037/a0016973.
[23] Cristianini, N., & Shawe-Taylor, J. (2000). An
introduction to support vector machines and
other kernel-based learning methods.
Cambridge university press. DOI:
10.1017/CBO9780511801389.
[24] Mohamed, S. M., & Nyongesa, H. (2002).
Automatic fingerprint classification system
using fuzzy neural techniques. IEEE World
Congress on Computational Intelligence.
2002 IEEE International Conference on Fuzzy
Systems, Honolulu, HI, USA. FUZZ-IEEE'02.
Proceedings (Cat. No. 02CH37291) (Vol. 1,
pp. 358-362). DOI:
10.1109/FUZZ.2002.1005016.
[25] Nur-A-Alam, Ahsan, M., Based, M. A.,
Haider, J., & Kowalski, M. (2021). An
intelligent system for automatic fingerprint
identification using feature fusion by Gabor
filter and deep learning. Computers and
Electrical Engineering, vol. 95, 107387. DOI:
https://doi.org/10.1016/j.compeleceng.2021.1
07387.
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
The authors equally contributed in the present
research, at all stages from the formulation of the
problem to the final findings and solution.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflicts of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on COMPUTERS
DOI: 10.37394/23205.2024.23.18
Md. Abdul Based, Elias Ur Rahman,
Mohammad Shorif Uddin
E-ISSN: 2224-2872
203
Volume 23, 2024