|
|
| |
|
Submit Paper / Call for Papers |
|
Journal receives papers in continuous flow and we will
consider articles from a wide range of Information Technology disciplines
encompassing the most basic research to the most innovative technologies. Please
submit your papers electronically to our submission system at
http://jatit.org/submit_paper.php in an MSWord, Pdf or compatible
format so that they may be evaluated for publication in the upcoming issue. This
journal uses a blinded review process; please remember to include all your
personal identifiable information in the manuscript before submitting it for
review, we will edit the necessary information at our side. Submissions to JATIT
should be full research / review papers (properly indicated in case of review papers). |
| |
Journal of
Theoretical and Applied Information Technology
July 2024 | Vol. 102 No.13 |
|
Title: |
SMOTE-2DCNN FOR ENHANCING SPEECH EMOTION RECOGNITION |
|
Author: |
NURUL NADHRAH KAMARUZAMAN, NOR AZURA HUSIN, NORWATI MUSTAPHA, RAZALI YAAKOB,
MUHAMMAD MUDASSIR EJAZ |
|
Abstract: |
Speech emotion recognition (SER) is a specialized form of audio classification
that aims to identify and classify emotional states expressed from spoken
language or speech signals. In this study, the main objective is to propose an
accurate audio classification model for the SER. This study primarily focuses on
two key issues: the insufficient training data within each available dataset and
the imbalanced distribution of data, both of which contribute to overfitting and
negatively impact the accuracy of the audio classification model. Henceforth, we
present the SMOTE-2DCNN, which is a combination of the Synthetic Minority
Oversampling Technique (SMOTE) with a 2-Dimensional Convolutional Neural Network
(2DCNN), designed to effectively address imbalanced data distributions and
achieve accurate emotion classification. Our proposed SMOTE-2DCNN demonstrates
outstanding performance with a UA rate of 81% and a WA rate of 80%. This
represents a substantial enhancement, achieving approximately 15% higher
accuracy compared to the leading state-of-the-art method. |
|
Keywords: |
Speech Emotion Recognition, Audio Classification, Deep Learning, SMOTE,
Imbalanced Data |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
5G APPLICATIONS VIA VIRTUAL REALITY TECHNOLOGY IN EDUCATION |
|
Author: |
ROSILAH HASSAN, MUHAMAD IRSAN, MIRNA NACHOUKI, NURUL HALIMATUL ASMAK ISMAIL,
SAMER ADNAN BANI AWWAD |
|
Abstract: |
Fifth-generation (5G) technology has been widely adopted in all spheres of
society, fostering excellent development across a range of sectors and domains.
In education, 5G technology has greatly improved the interactive communication
between teachers and students, and students and human-machine in the smart
teaching mode. With its functions for teaching, research, management, and
evaluation, the smart teaching mode has created a new paradigm for digital
education. It provides smart teaching cloud services to external tutors as well
as instructors and students at affiliated colleges and universities. However,
educational institutions today are still unaware of the importance of 5G and VR
(virtual reality) in education, because they do not apply their use in classroom
teaching and learning activities. In fact, they are still faced with unstable
network problems that interfere with the teaching and learning process.
Therefore, this preliminary study is dedicated to discussing the awareness of 5G
applications with VR technology in education. This is to see the extent of the
knowledge of instructors and students regarding the use of 5G and VR in their
educational activities. The study approach has been decided upon as an online
survey based on an opinion poll (questionnaire) due to the rapid turnaround,
prompt delivery, and simple return. The results showed that 90% of the
respondents said they had heard of VR technology, and 89.13% had used the 5G
application for teaching and learning. This shows that 5G technology has been
widely used in education, and VR technology is gradually entering people's
vision. In conclusion, this study will be able to give some awareness to
educational institutions in particular, to apply the use of 5G and VR in future
education. |
|
Keywords: |
5G, Education, Virtual Reality Technology |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
COGNITIVE FEATURES FOR EARLY ALZHEIMER'S DISEASE DETECTION: A STACKING-BASED
ENSEMBLE MACHINE LEARNING METHOD |
|
Author: |
PULI SUKESH, RADHIKA RANI CHINTALA |
|
Abstract: |
This work proposes a novel ensemble machine-learning approach for early AD
detection, focusing on cognitive features. The method employs a stacking-based
ensemble model, combining the strengths of multiple base learners to improve
prediction performance. It utilizes a comprehensive dataset containing cognitive
features, demographic information, and clinical scores from the Alzheimer's
Disease Neuroimaging Initiative (ADNI) database. The proposed method achieves
high accuracy in distinguishing between AD patients and healthy controls, with
several models, including Decision Tree, Decision Tree - NCA, Voting Classifier,
Voting Classifier - NCA, Stacking Classifier, and Stacking Classifier - NCA,
achieving 100% accuracy. This demonstrates the potential of the approach as a
valuable tool for early AD detection. A crucial advancement in this domain is
the adoption of ensemble machine learning models, which significantly enhance
the robustness of predictive systems by amalgamating diverse machine learning
algorithms. This novel approach incorporates a feature selection method referred
to as Neighborhood Component Analysis and Correlation-based Filtration (NCA-F)
to sift through a given dataset and pinpoint pivotal cognitive features. The
proposed research contributes to the advancement of early Alzheimer's disease
detection by leveraging machine learning techniques, specifically stacking-based
ensemble methods, to identify cognitive features indicative of the disease in
its early stages. |
|
Keywords: |
Alzheimer's disease, early detection, cognitive features, ensemble machine
learning, stacking, Alzheimer's Disease Neuroimaging Initiative (ADNI). |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
AUTOMATIC ROAD ROUGHNESS DETECTION AND RANKING USING DEEP LEARNING AND COMPUTER
VISION |
|
Author: |
MUHAMMED SAFFARINI, AMJAD RATTROOT, YOUSEF-AWWAD DARAGHMI, MUATH SABHA |
|
Abstract: |
Roads roughness is considered one of the most important problems that government
institutions face because it requires many complex issues to find the roughness
of the street. It also requires a lot of expensive tools which, in turn, measure
the roughness of the roads. so, in this research paper we create a new model
study road roughness and rank the roughness of this road automatically without
the need for any cost or human intervention. Our proposed model checks the
roughness by capturing the imaging using a drone, then it processes and analyzes
the images coming from the drone, using several models that work together. our
model shows the pattern of roads from the captured image using Gray level Size
Zone Matrix(GLSZM) features Zone Percentage (ZP) and Size Zone Non-Uniformity
(SZN) and then take the spikes of its distributions then take these spike to get
optimal value K for Kmean to segment the image, the result of first model enter
to second model that make sorting for this images depending on GLSZM features
(ZP and SZV) to improve the result of our model, after that the image enter to
CNN to get the outcomes by classifying it into which category this roughness
belongs. The best accuracy we achieved in our model reached 91.94%, which is a
very high accuracy, and therefore by a large percentage all correctly captured
images from the drone has accurate results. |
|
Keywords: |
Computer Vision, Deep Learning, GLSZM, CNN, Road Roughness |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
AN INTEGRATED ACCEPTANCE MODEL FOR VIRTUAL REALITY ADOPTION IN DISTANCE
LEARNING: INVESTIGATING MOROCCAN STUDENTS' PERSPECTIVES |
|
Author: |
LOBNA EL AMRANI, MOHAMED MOUGHIT |
|
Abstract: |
Virtual Reality (VR) is a technology with diverse applications across different
sectors, such as education, healthcare, psychology, and gaming. In education, VR
is being explored as a tool for distance learning. Its use can potentially
motivate students to engage with online lessons. The purpose of this research
paper was to investigate which variables would influence the use of VR in
distance learning among students. Using the Technology Acceptance Model (TAM) as
a framework within four factors, a series of hypotheses were formed. Data has
been collected from 122 Moroccan students and analyzed using regression. The
findings indicate that user support, perceived ease of use (PEOU), perceived
usefulness (PU), and attitudes toward technology use (ATU) significantly
influenced the behavioral intention (BI) to use VR systems for educational
purposes. The study's results can guide decision-makers in developing
sustainable distance learning and educational systems in Moroccan universities.
This study presents an integrated acceptance model for understanding Moroccan
students' perspectives on adopting VR in distance learning. By investigating
factors influencing students' acceptance of VR technology, this research
contributes to the development of sustainable distance learning systems in
Moroccan universities |
|
Keywords: |
Distance education, Virtual Reality, Technology Acceptance Model, e-learning,
TAM |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
E-LEARNING USING ARTIFICIAL INTELLIGENCE: AN INNOVATIVE APPROACH TO DISTANCE
LEARNING FOR ENHANCED DATA GENERATION |
|
Author: |
A. LECHHAB, M. EZZAKI, D. BENAMMI, M. BOUJARRA1, Y. FAKHRI, S. BOUREKKADI |
|
Abstract: |
The article highlights the increasing development of intelligent systems in the
modern world, aimed at simplifying human learning. It highlighted the emergence
of e-learning, designed to disseminate knowledge using artificial intelligence
as a solution for higher level education. The main goal of e-learning is to
deliver high-quality education in an efficient manner, based on sound
technological design. Creating e-learning courses is presented as a complex and
expensive task, involving many people and skills. However, recent advances in
artificial intelligence offer the possibility of automating this process. This
article proposes an innovative approach to strengthen data security in the field
of e-learning by integrating artificial intelligence (AI). Using advanced data
analysis and statistical modeling techniques, we identify potential
vulnerabilities and propose proactive measures to mitigate risks. Our method
uses AI to monitor suspicious activities in real time and adapt security
policies accordingly. By leveraging AI's versatility in anomaly detection and
malicious behavior prediction, our approach provides dynamic defense against
emerging threats. The results of this study demonstrate the effectiveness of
AI-powered e-learning in ensuring data security while optimizing remote learning
processes. The authors thus propose the automatic generation of e-learning
courses using intelligent systems, claiming that this method would be more
effective than traditional course development methods. The study focuses on
automatic generation of e-learning courses, followed by evaluation using concept
maps. The researchers claim that this approach is not only more effective than
traditional methods, but also that the quality of the courses generated is
higher. The article highlights the potential of artificial intelligence to
transform the way e-learning courses are developed and delivered, providing a
more efficient and higher quality solution for online education. |
|
Keywords: |
E-learning, Artificial intelligence, Automatic course generation, Concept maps |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
ENHANCING RETAIL PRODUCT RECOGNITION USING MODIFIED YOLOV8 AND SELF-SUPERVISED
LEARNING |
|
Author: |
FELIX CORPUTTY, SURYO ADHI, WIBOWO, UNANG SUNARYA3, RISSA RAHMANIA, SIDDIQ WAHYU
HIDAYAT |
|
Abstract: |
Artificial intelligence has several parts, one of them is computer vision.
Computer vision is a technology that allows computers to recognize objects as
humans do. Computer vision has been widely applied in various applications, one
of an example is in retail product recognition. However, the current computer
vision technology is still difficult to distinguish between one product and
another in the same category known as intra-class variation. Therefore, this
research developed an algorithm that uses the concept of computer vision to be
able to distinguish one product and another in the same category. The research
was conducted using two stages. In the first stage the dataset was trained using
YOLOv8. There are four experiments conducted using YOLOv8, namely YOLOv8
original, YOLOv8 with 4 detection heads (YOLOv8-4DH), YOLOv8 with additional
convolutional layer and C2f layer on the backbone (YOLOv8-Conv) and the last is
YOLOv8 with 4 detection heads, additional convolutional layer and C2f layer on
the backbone (YOLOv8-Conv-4DH). The best model is selected based on the highest
mAP value. The model with the highest mAP value is YOLOv8-4DH at 91%. The best
model is used to crop the image to be used as input in the second stage. In the
second stage, the cropped image is trained using SimCLR. The training weights
from SimCLR are stored and loaded back into the SimCLR model for training and
evaluation. The results of the second stage showed that the best model
YOLOv8-4DH combined with SimCLR algorithm got an accuracy of 97.76%. |
|
Keywords: |
Artificial Intelligence, Computer Vision, Retail Product, SimCLR, YOLOv8 |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
DETECTION OF GEOMAGNETIC STORM SUDDEN COMMENCEMENTS WITH THE USE OF NEURAL
NETWORK ARCHITECTURES |
|
Author: |
TARAS VOLOSHIN, KONSTANTIN ZAYTSEV |
|
Abstract: |
The purpose of the study is to examine various options to address the task of
detecting the starting stage of geomagnetic storms, storm sudden commencement
(SSC or SC), based on measurements of the Earth's magnetic field collected by
INTERMAGNET observatories. These observatories are located in different regions
of the world, allowing the full range of geomagnetic observations to be
processed. Through a comprehensive analysis involving time series and machine
learning techniques, including both statistical and neural network models, we
developed models that integrate scalar and vector data to enhance detection
accuracy. Discontinuities on the time scale in the measurements of individual
observatories have been registered. In addition to the time series of magnetic
field measurements, sudden commencement was detected using such scalar values as
the change of the level of induction components and change of rhythm. Various
methods of modeling and analyzing time series have been proposed, including
statistical and machine-learning methods. To use vector and scalar indicators at
the same time, the model was built with two streams of information. Various
models were built using the data of both single and multiple laboratories. In
the latter case, data from different sources were combined by the methods of
hard voting and soft voting. A quantitative assessment of the results delivered
by the models was carried out using accuracy, recall, and precision metrics. |
|
Keywords: |
Geomagnetic Storms, Sudden Commencement, Time Series, Machine Learning, Neural
Networks. |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
ELECTROCARDIOGRAM PLETHYSMOGRAPHIC ELECTROMYOGRAMS BASED BIOMETRIC
AUTHENTICATION MODELS |
|
Author: |
SUNEETHA MADDULURI, T. KISHORE KUMAR |
|
Abstract: |
To verify humans identity, biometric authentication techniques examine
observable characteristics. This may be based on a person's fingerprint, iris,
retina, Electrocardiogram (ECG), Plethysmographic (PPG), Electromyograms (EMGs)
or some other identifying features. There is flexibility in the usage of a
single trait or a combination of traits. Because they are both discrete and
distinctive, electrocardiograms (ECGs), photoplethysmograms (PPGs), and
electromyograms (EMGs) have been investigated as possible biometric features in
the last several decades. Research into biometric recognition technologies that
are user-unobtrusive has been accelerated by the increased availability of
wearable sensors and mobile devices. Due to their distinct characteristics,
electrocardiogram (ECG) signals have recently been investigated as a potential
biometric identification trait. An electrocardiogram (ECG) can only be used to
collect data from individuals who are still alive, as it measures the electrical
activity of the heart. The research community is interested in evaluating
cardiac signals derived from PPG signals for a number of reasons, one of which
is the capacity to perform continuous authentications with affordable devices
that can gather signals without user intervention. With the declining quality
and resolution of gathered images and security issues such spoofing and copying,
this study intends to discuss and analyze biosignals based biometric
authentication, which has been dominating former conventional methods. This
research provides a brief analysis of ECG, PPG and PCG and their advantages and
limitations and proposed an ECG based Biometric Authentication using CNN
(ECG-BA-CNN). This analysis helps numerous researchers to design novel biometric
innovations overcoming the limitations of traditional models. |
|
Keywords: |
Electrocardiogram, Plethysmographic, Electromyograms, Biometric Authentication,
User Identity, Security. |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
THE IMPACT OF ARTIFICIAL INTELLIGENCE ON IMPROVING TEXT IN THE PROCESS OF
CONCEPTUALIZATION IN BIOLOGY: CASE OF EDUCATION SECTOR |
|
Author: |
GHIZLANE GHARIZ, HAKIMA SEGHIR, NAJAT BOUCETTA, SAID BOUBIH, RACHID
JANATI-IDRISSI, MUSTAFA EL ALAOUI |
|
Abstract: |
This article investigates the influence of artificial intelligence (AI) on
enhancing the process of text production, emphasizing the advantages,
difficulties, and consequences associated with its use. The author emphasizes
the transformative impact of sophisticated language models on the development of
textual material, since they provide high-quality output that is both natural
and instructive. The use of artificial intelligence (AI) has shown a significant
enhancement in productivity and a reduction in production time. However, it is
crucial to acknowledge that the integration of AI also presents ethical dilemmas
that need meticulous examination and contemplation. This article explores the
impact of artificial intelligence (AI) on written communication, focusing on its
influence on contextual comprehension, creative enhancement, and the
transformation of human linguistic interactions and perspectives. Furthermore,
the paper delves into contemporary implementations of artificial intelligence
(AI), including automated writing, chatbot systems, and educational contexts.
The study ultimately delves into the integration of artificial intelligence in
the creative process, specifically focusing on co-creation, and also explores
the reinterpretation of literary genres. Although AI offers several advantages,
it also presents ethical dilemmas, including those related to data bias and
editorial accountability. The promise of AI-assisted text production in the
future seems great; yet, its successful implementation requires continuous
ethical oversight and a comprehensive comprehension of the associated
ramifications. This approach is crucial in order to optimize the advantages
while effectively addressing any possible obstacles that may arise. |
|
Keywords: |
Artificial Intelligence, Text Generation, Linguistic Models, Editorial
Creativity |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
CYBERSECURITY RISK MANAGEMENT IN IOT SYSTEMS: A SYSTEMATIC REVIEW |
|
Author: |
TAYSEER ALKHDOUR, MOHAMMED AMIN ALMAIAH, MARIAM ALI ALAHMED, MOHMOOD A.
AL-SHAREEDA, ABDALWALI LUTFI, MAHMAOD ALRAWAD |
|
Abstract: |
With the revolution of IoT technologies, cybersecurity risks are considered one
of the challenges of IoT. Therefore, this study aims to discuss the risk
management process for IoT in order to identify the main vulnerabilities and
threats in IoT. In addition, this paper discusses the best mitigation techniques
and risk management frameworks and models in order to ensure that the IoT users
protected from any cyber-attacks. The study indicates that the DDoS attacks is
the highest percentage of risk in IoT technologies. The paper also finds that
IoT risks can be divided into four types including privacy risks, security
risks, technical risks and ethical risks. The study find that the ISO is the
best framework for the risk management in IoT technologies. Finally, the paper
presents for researchers important recommendations for determining the types of
risks and attacks in IoT and identifying the most important risk management
frameworks and models for IoT. |
|
Keywords: |
IoT; Privacy; Cybersecurity Risks; Cybersecurity Management; User Privacy;
Blockchain. |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
VOCATIONAL EDUCATION SKILL ASSESSMENT AND INTELLIGENT ASSISTANCE: A STUDY ON THE
APPLICATION OF MACHINE LEARNING ALGORITHMS IN THE ASSESSMENT OF VOCATIONAL
INFORMATION LITERACY TEACHING ABILITY |
|
Author: |
Dr. Vishal M. Tidake, Dr. Utpala Das, Dr. Kulbir Kaur Bhatti, R. Swathi
Gudipati, Dr. S. Farhad, Prof. Ts. Dr. Yousef A.Baker El-Ebiary, Manikandan
Rengarajan |
|
Abstract: |
Vocational education skill assessment and intelligent assistance involve
evaluating people' talent in precise vocational abilities and conveying
personalized assist to enhance studying results. The need for such assessment
and assistance arises from the significance of appropriately evaluating
learners’ readiness and proficiency in vocational abilities, identifying areas
for improvement in teaching practices, and presenting timely feedback and
guidance to learners. However, present strategies often depend on conventional
assessment techniques which can lack granularity and fail to provide
personalised assistance. To address those demanding situations, this study
introduces a novel method that integrates SMOTE data processing, Federated LSTM
(Fed-LSTM) for skill word extraction and classification, and fuzzy rule-based
vocational education talent evaluation. This approach targets to overcome class
imbalances in datasets through SMOTE, permit collaborative learning across
distributed data sources, and improve the accuracy and robustness of talent
assessment models. The proposed study improves data representation, facilitating
collaborative learning, enhancing skill extraction accuracy, and presenting
robust skill assessment. The results of study are applied in a Python software,
offering educators and stakeholders a realistic approach to enhance vocational
education skill assessment and intelligent assistance. The proposed Fed-LSTM
technique demonstrates a substantial growth in accuracy compared to the LSTM
approach. With an accuracy of 99.4%, the proposed technique considerably
outperforms the LSTM method, which achieved an accuracy of 76. 98%. This
represents a substantial improvement of 22.42% in accuracy. |
|
Keywords: |
Vocational Education, Skill Assessment, Intelligent Assistance, Teaching
Practices, Synthetic Minority Oversampling Technique |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
AN EFFICIENT AND ROBUST PROOF OF STAKE ALGORITHM BASED ON COIN-AGE SELECTION |
|
Author: |
ANJANEYULU ENDURTHI, AKHIL KHARE |
|
Abstract: |
A consensus protocol is used to achieve agreement among the nodes in a
distributed system. Proof of stake is one such protocol. Proof of stake is based
upon two different strategies. The first one is randomized block selection and
the second is coin-age selection. Each of these strategies results in an unfair
selection of validators and converges to a problem called wealth concentration
among a few validators. This paper proposes a modified proof of stake protocol
based on the coin-age strategy to mitigate the issue and improve the coin-age
selection algorithm. The participants will generate new tokens to compete for
the validator role to create the next block. |
|
Keywords: |
Blockchain, Consensus, Proof of stake, Coin-age, Timestamp, Tokens |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
DESIGN OF AN ITERATIVE METHOD FOR PLANT NUTRIENT DEFICIENCY DETECTION USING
GRAPH CONVOLUTIONAL NETWORKS & ENSEMBLE LEARNING |
|
Author: |
PARNAL P. PAWADE, DR. A. S. ALVI |
|
Abstract: |
Identifying nutrient deficiencies in plants with enhanced precision is crucial
for sustainable food production. Traditional methods often fail to capture the
complex biological scenes in various use cases. This work introduces a novel,
precision-aware, learning-based approach to significantly improve the detection
and classification of nutrient deficiencies in plants. Unlike available
methodologies that rely solely on image-based analysis, our method employs Graph
Convolutional Networks (GCNs) to create graph-based representations of plant
structures from high-resolution images. This technique captures intricate
relationships between plant parts, such as leaves, stems, and roots, by treating
them as interconnected nodes in a graph. GCNs extract hierarchical features,
providing a comprehensive and discriminative representation for nutrient
deficiency detection. We also propose an ensemble model combining Capsule
Networks and Transformers. Capsule Networks understand hierarchical and spatial
relationships within plant data, while Transformers capture long-range
dependencies and complex patterns across various plant sections. This
combination results in an ensemble with enhanced accuracy. To overcome the
limitations of training data and biases in real samples, we introduce a novel
data augmentation method using Generative Adversarial Networks (GANs). This
method generates synthetic images reflecting real growth variations, lighting
conditions, and nutrient deficiency symptoms, thus improving model
generalization and robustness. Furthermore, we present an innovative
interpretability technique to display attribution-based visualizations of
graph-based features. This approach elucidates the model's reasoning by
identifying influential regions and structures within the dataset, thereby
increasing trust in the model's decisions and providing biologically relevant
insights. Our method advances agricultural technology by enhancing nutrient
deficiency detection accuracy and interpretability, aligning with biological
agricultural knowledge. This comprehensive approach paves the way for more
sustainable and informed agricultural practices, leading to improved crop health
and productivity. |
|
Keywords: |
Graph Convolutional Networks, Ensemble Learning, Plant Nutrient Deficiency, Data
Augmentation, Interpretability Techniques |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
A MACHINE LEARNING-BASED OPTIMIZED FRAMEWORK FOR DETECTION OF ARRHYTHMIA FROM
ECG DATA |
|
Author: |
MS.K.SHILPA, DR.T.ADILAKSHMI |
|
Abstract: |
Heart diseases are causing health issues for people across the globe due to
various reasons, including lifestyle changes. With the emergence of artificial
intelligence (AI), it is possible to have learning-based approaches for the
automatic detection of several types of heart diseases. Many existing approaches
followed data-driven techniques for the diagnosis of heart diseases. Some
methods focused on ECG data, which has the potential to support the detection of
different kinds of diseases, particularly arrhythmia. The literature shows that
machine learning models result in deteriorated performance unless specific
optimizations support them. Motivated by this fact, we proposed a
machine-learning framework that exploits many classification models for
detecting arrhythmia and classification. The proposed framework is subjected to
multiple optimizations in terms of preprocessing, feature engineering, and
hyperparameter tuning. To develop an optimized machine learning approach, we
proposed two algorithms known as Feature Selection and Hyperparameter
Optimization (FSHO) and Learning-based Arrhythmia Detection and Classification
(LbADC). We used our empirical study's benchmark dataset, known as the MIT-BIH
Arrhythmia dataset. The experimental results reveal that the proposed
optimizations and machine learning framework could improve arrhythmia diagnosis
and classification performance. The proposed optimizations of our framework
achieved 96.8% accuracy in multi-class classification. |
|
Keywords: |
Healthcare, Machine Learning, Feature Engineering, Heart Disease Prediction,
Arrhythmia Diagnosis |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
EXPLORING EMERGING CYBERSECURITY RISKS FROM AI-BASED IOT CONNECTIONS |
|
Author: |
HANAN KHALID ALSUWAELIM |
|
Abstract: |
The Internet of Things connects things and networks, such as devices and
infrastructure, in an at-tempt to make life easier. These networked areas,
however, frequently have few resources and are thus the most susceptible to
assaults. We must search for an all-encompassing security strategy for the
Internet of Things that safeguards these nodes as well as the data they manage.
In addition to the existing security protocols for networks, we might employ
intelligent strategies deriving from artificial intelligence principles and
basic and sophisticated machine learning approaches to pre-vent threats. The
future may be brighter if artificial intelligence is connected to the Internet
of Things. The aim of this paper is to review and analyze the cyber risks and
threats associated with IoT devices and artificial intelligence published from
2020 to 2024. Then, the paper highlights privacy and ethical concerns,
introduces security frameworks and tactics, classifies IoT security
difficulties, explores the use of AI-based in IoT security, and offers insights
from real-world case studies. A total of 25 articles were selected using the
PRISMA framework. This thorough analysis of the status of IoT security today and
how AI affects it advances our knowledge of how to create trustworthy and secure
IoT systems. |
|
Keywords: |
Intelligence, Security, Risk, Risk Analysis, and Internet of Things. |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
ISSUES AND CHALLENGES IN ONLINE LEARNING: A CASE STUDY IN MALAYSIA |
|
Author: |
FAAIZAH SHAHBODIN, ZULISMAN MAKSOM, CHE KU NURAINI CHE KU MOHD, HANIF AL-FATTA,
ULKA CHANDINI, HELMI MOHD KASIM |
|
Abstract: |
This paper focused on exploring the secondary school teachers’ perceptions
toward online learning program which developed during COVID-19 pandemic in
Malaysia. To evaluate teachers’ perceptions of teaching and learning engagement,
a quantitative survey was conducted. Therefore, teachers’ understanding of
teaching and the relation to their engagement in learning are explored in this
survey. Hence, there are factors which determined the success of implementing
online learning in Malaysia during COVID-19 pandemic such as the readiness of
technology which in line with the national humanist curriculum, support and
collaboration from all stakeholders including government, teachers, parents,
schools, and community. The findings in this paper highlight the teachers’ good
sense of teaching and strong correlations between teachers’ perceptions and
students’ engagement are significantly influence the online teaching and
learning process. Teachers are also suggested to apply more appropriate types of
learning tools during classes and pay attention to the nature of the student.
The results may assist in advocating for a paradigm shift in online education.
The research underscores the need for innovative approaches that leverage the
power of technology to inspire online learners and educators thereby
contributing to the ongoing improvement of online education. |
|
Keywords: |
Technology, Personalize Learning, Online Learning, Higher Education |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
ENHANCING SECURE V2V AND V2I COMMUNICATION: DESIGNING AN EFFICIENT
VEHICLE-AUTHORIZED SHORTEST ROUTE SELECTION ALGORITHM FOR MINIMIZING DATA LOSS |
|
Author: |
SPANDANA MANDE, NANDHA KUMAR RAMACHANDRAN |
|
Abstract: |
Ensuring the security and efficiency of information exchange between vehicles
(V2V) and infrastructure (V2I) is of utmost importance in the realm of vehicular
communication systems. The main objective of this study is to enhance the
security and efficiency of these systems. To achieve this, we will create a
vehicle-authorized algorithm for selecting the shortest route, to minimize data
loss. This algorithm employs cryptographic authentication mechanisms to
prioritize secure routes based on vehicle authorization, effectively mitigating
potential security risks during information exchange. Implementing optimized
routing protocols, such as the A* algorithm, allows the system to determine the
most efficient routes for vehicles, taking into account factors like traffic
conditions and network congestion. It specifically focuses on the critical
issues of choosing the best route and ensuring data security in communications
between vehicles. This algorithmic solution enhances both route selection and
network security while also establishing a robust framework for secure and
efficient vehicular communication systems. It guarantees the accurate and secure
reception of information between vehicles or infrastructure. |
|
Keywords: |
Vehicle to Vehicle, Vehicle to Infrastructure, A *, Shortest Route Selection,
Vehicular Communications, Data Loss. |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
ENHANCING GLAUCOMA DIAGNOSIS: DEEP LEARNING MODELS FOR AUTOMATED IDENTIFICATION
AND EXPLAINABILITY USING FUNDUS IMAGES |
|
Author: |
MANEESHA VADDURI, KUPPUSAMY. P |
|
Abstract: |
Glaucoma is a serious eye condition that poses a significant threat to vision
health, often resulting in permanent sight loss by damaging the optic nerves.
Detecting glaucoma early is crucial for effective management, aiming to reduce
intraocular pressure and inflammation. However, current detection methods are
resource-intensive and prone to human error, failing to detect the disease in
its early stages. Deep Learning (DL) offers promising avenues for automated
diagnosis, yet concerns persist regarding model reliability. Addressing this,
the Enhanced Deep Learning Approach for Glaucoma Diagnosis (EDAGD) is
introduced. Leveraging SegNet and ResNet-50 architectures, EDAGD achieves
exceptional segmentation accuracies of 98.58% for the Optic Disc (OD) and 96.52%
for the Optic Cup (OC) on the RIM-ONE dataset, while also demonstrating robust
performance on the ACRIMA and REFUGE datasets. Furthermore, EDAGD utilizes
cutting-edge visualization techniques such as Gradient-weighted Class Activation
Mapping (Grad-CAM) and Grad-CAM++ to generate interpretable heatmaps, aiding in
pinpointing critical regions for diagnosis. By accurately classifying segmented
images, EDAGD achieves impressive performance metrics of 97.97% accuracy, 98.41%
sensitivity, and 96.58% specificity. The potential impact of automated glaucoma
diagnosis on healthcare systems includes reducing the burden on
ophthalmologists, increasing accessibility to diagnostic tools in remote areas,
and potentially lowering healthcare costs. By integrating advanced Deep Learning
techniques with explainable AI methods, our approach not only improves the
accuracy of glaucoma diagnosis but also builds trust among clinicians. This
fosters seamless integration into clinical practice, ultimately advancing
patient care by enabling timely and accurate diagnosis of glaucoma. |
|
Keywords: |
Glaucoma, segmentation, classification, Fundus images, Explainability |
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
Title: |
EVALUATING TEXTBLOB, LEXICON, SUPPORT VECTOR MACHINE, NAIVE BAYES, AND CHATGPT
APPROACHES FOR SENTIMENT ANALYSIS OF NASDAQ LISTED COMPANIES |
|
Author: |
AZWARNI, NATHAR SHAH |
|
Abstract: |
Sentiment analysis is a type of contextual text mining that finds and extracts
subjective information from the source material in order to assist companies in
understanding the social sentiment of their brand, product, or service while
monitoring online conversations, especially Twitter has become a popular medium
for individuals to express their opinions, share news, and discuss various
topics, including stocks and companies. Stock market sentiment analysis is
useful for understanding investor sentiments and forecasting market moves.
Market players can use sentiment analysis tools to evaluate market sentiment and
make educated investing decisions. The previous study examined data with fewer
than ten thousand rows; however, this research will work with very huge data
sets of more than one hundred thousand Nasdaq companies acquired from @Nasdaq
and @AppleSupport Twitter accounts and @nasdaq and @apple from subreddit in
Reddit social media. This study will compare the classification accuracy of
Naive Bayes and SVM, as well as the time consumption of each strategy while
classifying vast quantities of data. The TextBlob NLTK (Natural Language
Toolkit) will be used in this study to label each phrase in the data using a
lexicon-based method; also, this effort will employ ChatGPT, an OpenAI chatbot,
to label each row of data received. As a consequence, it was discovered that SVM
is the most superior approach in its classification, both in terms of Precision,
Recall, and F1-Score metrics, as well as total accuracy, which reaches 93.5%,
while Naive Bayes is at 61.5% and ChatGPT is at 42.2%. |
|
Keywords: |
Big Data, TextBlob, SVM (Support Vector Machine), Naïve Bayes, ChatGPT,
Sentiment Analysis, Nasdaq
|
|
Source: |
Journal of Theoretical and Applied Information Technology
15th July 2024 -- Vol. 102. No. 13-- 2024 |
|
Full
Text |
|
|
|
