International Journal of Computer and Information Technology
Not a member yet
138 research outputs found
Sort by
Enhancing Image Processing Capabilities based on Optimized Neural Networks: Image identification and classification
No full textImage processing is the ability of machines to interpret and understand visual data, has been significantly advanced by Convolutional Neural Networks (CNNs). This study investigates the enhancement of image procesing performance through the optimization of CNN architectures. By performing comparison between basic CNN models with optimized versions, incorporating advanced techniques such as deeper convolutional layers, batch normalization, dropout, and data augmentation, the aim of the study is to improve accuracy and robustness in image detection and classification tasks. The experiments are carried out on benchmark datasets and the results demonstrate that optimized CNNs substantially outperform their basic counterparts, achieving higher training and validation accuracies. These findings highlight the critical role of architectural refinements and regularization techniques in advancing visual intelligence capabilities. This research presents a novel approach that underscores the capability of optimized CNNs to drive future innovations in the area of visual intelligence, offering more accurate and reliable visual data interpretation for real life applications
Detailed Analyses and Efficient Identification of Malware Evidence in CLaMP Dataset based on Machine Learning Approaches
Get PDFMalware is a malicious software that is used to launch attacks of different types in computer networks and cyber space. Several signature and machine learning-based approaches have been used for the identification of malware types in the past. However,signature-based detection approaches have been reported to have serious limitations which gave room for machine learning-based malware identification techniques to be more popular. Despite the promises of the ML methods in the identification of malware evidence, some of the ML approaches in literature have poor detection rates which can be as a result of the size and nature of the patterns in the datasets used. This study used a dataset named CLaMP for the training and testing of the malware classification models. Firstly, comprehensive exploratory analyses of the dataset were carried out with a view to understanding the data distributions in it better and make informative decisions on how to pre-process and apply it for malware identification. During the experimentations, two scenarios were established before feeding the data into the learning algorithms. Scenario 1 involves building malware identification model without data cleaning and feature selection while scenario 2 involves the cleaning of the data and selection of promising features for building the models.In scenario 2, Recursive Feature Elimination (RFE) technique was used for selecting the promising attributes which were used to build the two malware classification models. Naive Bayes (NB) and Logistic Regression (LR) algorithms were used for building the models. The hyper parameters of the two selected algorithms were varied and the models tested and validated severally before optimal performances were arrived at. The results of the models were compared based on the selected metrics, namely: accuracy, precision, recall, f1-score and Area Under the Curve (AUC). Experimental results showed that in the scenario 1, where the dataset was not pre-processed and all the attributes were used for the model building, poor results were obtained by both models in all metrics except in recall. However, NB-based malware identification model slightly performed better than LR in all the metrics. It was also discovered that both NB and LR-based malware identification models performed well in scenario 2 when the dataset was pre-processed and promising features were selected using RFE. This study concluded that the detailed exploratory analyses, data cleaning and feature subset selection methods helped in achieving promising results from the malware identification models
Features of Lightweight Proof of Stake Models for Enhancing Data Privacy in Telemedicine Systems: A Systematic Literature Review
No full textProof of Stake (PoS) models are energy-efficient and require limited computational power. These features are critical in telemedicine environments, where resource-constrained devices must handle sensitive data securely. The growing need for auditable and privacy-preserving data storage in telemedicine underscores the importance of PoS models optimized for lightweight devices while complying with strict regulatory requirements, such as the Health Insurance Portability and Accountability Act (HIPAA).This study was guided by two research questions: (i) Which PoS models are lightweight and suitable for telemedicine? and (ii) What features make lightweight PoS models effective for privacy and efficiency in telemedicine? To address these questions, a systematic literature review (SLR) guided by the PICOC framework was conducted to investigate lightweight PoS models that can enhance privacy in telemedicine systems. Out of 2,394 papers studies screened, 55 were included in the analysis. The findings identified Algorand, Ouroboros Praos, Tendermint, Nxt, and Casper CBC as promising candidates. Key enabling features included lightweight voting mechanisms, such as Byzantine Agreement protocols and Verifiable Random Functions, as well as cryptographic techniques like symmetric encryption and multiparty computation. Performance metrics evaluated included latency, throughput, energy efficiency, and battery consumption, with Grey Relational Analysis ranking Algorand highest due to its low latency, high throughput, and minimal energy consumption
The Assessment of Organizations Readiness to Comply with the Kenya Data Protection Act: A Case of a Humanitarian Arganization
Get PDFPrivacy is a crucial aspect of life as it impacts on how we behave, feel and make decisions. It recognizes the dignity and inherent worth of individuals. The right to privacy as a fundamental right is recognized in our 2010 constitution under article 31 sub article c & d. Kenya enacted the Kenya Data Protection Act in 2019 (KDPA, 2019) to safeguard personal information, in accordance with a set of statutory principles.
The act requires organizations to register with data commissioner’s office (ODPC), demonstrate safeguards in place for personal data processing, carry out a data protection impact assessment (DPIA) for processes that pose a significant risk to the privileges and autonomies of its citizens and report any breach within 72 hours.
In order to evaluate an organization\u27s compliance with the act, it is imperative to perform readiness assessment to review organizations privacy practices across different domains and identify any gaps as well as the necessary steps for achieving and maintaining compliance.
To streamline readiness evaluation therefore, this study reviewed the privacy maturity models currently in use for organizations to measure their readiness to comply with privacy laws and assessed readiness of a humanitarian organization to comply with the act.
The AICPA / CICA privacy maturity model informed the readiness assessment of the humanitarian organization to KDPA compliance. The study adopted Quantitative research methodology.
The research identified regulatory, culture and technology readiness as dimensions influencing organizations readiness to comply with KDPA and to improve the overall readiness score, organizations need to put emphasis on all the three domains (regulatory, culture and technology).
Organizations can evaluate their compliance with the provisions of the act using the study’s findings, identify areas of non-compliance and prioritize remediation efforts
Big Data Analytics in Healthcare: Predictive Modeling, Privacy Challenges, and Global Regulatory Compliance
Get PDFBig Data Analytics (BDA) is increasingly central to modern healthcare, promising transformative improvements in patient care, operational efficiency, and predictive disease modeling. However, the sensitive nature of health data also introduces significant challenges, particularly regarding privacy, confidentiality, and global regulatory compliance. This article synthesizes key insights from contemporary research, highlighting methods, existing gaps, proposed solutions, and the critical need for stronger global data privacy harmonization
Topics, Trends, and Sentiments in Software Testing: An Analysis of Developers’ Engagement on Stack Overflow
Get PDFThis study investigated software testing discussions on Stack Overflow from 2020 to 2024 to uncover key trends, topics, and developer sentiments. 14 key topics, including unit testing, machine learning testing, mobile testing (especially Flutter), and Docker testing were identified. The study revealed a decline in developer engagement, as the number of posts answered and with accepted answers decreased, particularly after 2022. Sentiment analysis showed a predominance of negative sentiments across most topics, especially in mobile and machine learning testing. While some topics like machine learning testing initially saw positive sentiment, this shifted toward frustration as the years progressed. These findings suggest that the rise of AI-based tools, such as ChatGPT, has affected the way developers engage with traditional forums like Stack Overflow. The decline in engagement and the prevalence of negative sentiments highlight the challenges developers face in software testing. This research points to the need for further investigation into how AI tools influence developer behavior and their reliance on peer support platforms. Additionally, it suggests exploring how sentiment analysis can be integrated into software testing tools to better address developer frustrations and improve support for testing emerging technologies. The study provides insights that could guide the development of more effective tools and frameworks to enhance the software testing process
A Centrality Maximization Approach for Link Recommendation
Get PDFIn social networks, the goal of link recommendation is to recommend links for nodes and add them to the network, thereby satisfying the potential link interests of the nodes. The centrality of nodes in social networks typically quantifies the importance of nodes in the network. Some nodes may desire to increase their centrality by adding links. First, a multi-community centrality measurement method is proposed, and based on harmonic centrality, a hybrid centrality measurement method is introduced. Next, the link recommendation problem is regarded as a problem of maximizing node hybrid centrality, which can be formally modeled as a submodular function maximization problem. A greedy algorithm with performance guarantees can be directly applied to select the best links. Compared to existing link prediction and link recommendation algorithms, our algorithm recommends links that better improve the hybrid centrality of users
A SYSTEMATIC LITERATURE REVIEW ON PHISHING DETECTION MODEL
Get PDFThis paper introduces a unique method using supervised learning techniques in a hybrid crime detection model to identify phishing attempts on social media sites. Effective detection systems are desperately needed given the rise in criminality on social media, especially phishing. The suggested model combines the best features of several supervised learning algorithms which comprises of random forest, decision tree, support vector machine which are frequently used in analyzing the phishing attacks, taking use of their capacity to extrapolate patterns from labeled datasets and spot questionable behavior suggestive of phishing efforts. The commonly used algorithm was Decision Tree (DT), with 14% of the total, followed by Random Forest (RF), Support Vector Machine (SVM), and Naïve Bayes (12%), with 8%. The least popular algorithms were LSTM, SCS, STARMA, AUC, and FURIA, with 2% each.
Decision trees and Support Vector Machines (SVMs) are often used in phishing assault detection since they excel at classification tasks exactly what phishing detection entail. The reason for this is their ability to differentiate between trustworthy and malevolent websites or emails. Decision trees offer a clear and concise example of decision-making processes.
Decision tree (DT) presents several gaps which need to be solved, should important characteristics associated with phishing offenses be omitted or misidentified, the efficacy of the model may be jeopardized. Overfitting and class imbalance is a common problem with decision trees, particularly when working with complicated datasets. This might result in poor generalization to fresh, untested data, which would make the model less effective at identifying unusual phishing scams. Phishing statistics on social media frequently exhibit a class imbalance, with a comparatively smaller number of phishing crimes than lawful activity
Assessing IPv6 Readiness and Adoption in Palestine, A Case Study with Strategic Recommendations
Get PDFThis research examines the current state of IPv6 adoption in Palestine, comparing it to the global and regional adoption trends, particularly in Arab countries. While the global IPv6 adoption rate exceeds 45%, and Arab countries average 12.79%, Palestine remains at a negligible adoption rate. Through a mixed-method approach that includes experiments, technical analysis, and stakeholder interviews, the study highlights the readiness of end-user devices for IPv6, primarily due to international compliance standards. However, significant challenges exist within Internet Service Providers (ISPs) and large enterprises in Palestine, such as dual-stack resource demands, insufficient technical expertise, and lack of customer demand. This paper underscores the urgent need for a national strategy, stakeholder collaboration, and increased awareness to accelerate IPv6 transition in Palestine. Recommendations are made to address these obstacles and position Palestine for future compatibility in the global internet ecosystem
The Trade-Off Between Anonymity and Accountability in Blockchain: A Framework for Secure and Compliant Systems
Get PDFBlockchain technology has revolutionized digital transactions by offering decentralization, transparency, and immutability. However, its inherent transparency often conflicts with the need for user privacy and anonymity, raising significant concerns regarding accountability, especially in regulatory and legal contexts. This study explores the delicate balance between anonymity and accountability in blockchain systems, proposing a framework that ensures both privacy and compliance with regulatory requirements. The research addresses key challenges in balancing these two aspects, evaluates the effectiveness of existing privacy-preserving technologies such as zero-knowledge proofs and ring signatures, and introduces the Privacy-Accountability Balanced Blockchain (PABB) Framework. This framework integrates Selective De-Anonymization, Self-Sovereign Identity (SSI), and the Adaptive Privacy-Accountability Control (APAC) Algorithm to dynamically adjust privacy levels based on regulatory conditions. Through theoretical analysis, mathematical modeling, and empirical validation, preserving privacy for 92% of transactions while enabling selective de-anonymization in high-risk cases, the study demonstrates that the APAC Algorithm effectively balances privacy and compliance needs. The findings suggest that privacy-conscious blockchain systems can coexist with accountability mechanisms, paving the way for ethical and legally sound blockchain applications. The study concludes that the PABB Framework offers a practical and scalable approach to achieving this balance, fostering trust among users and regulators alike