Asian Journal of Research in Computer Science
Not a member yet
792 research outputs found
Sort by
Predicting Customer Churn in Telecommunications with Machine Learning Models
No full textCustomer churn is an important issue in businesses and service sectors including telecommunication industries. Prediction of potential customer churn can be very useful in these fields and it can help to improve customer retention significantly by providing personalized offering to reduce potential churn. This paper mainly focused on customer churn prediction using machine learning (ML) models and Iranian Telco customer churn dataset. Different reasons or variables are involved in customer dissatisfaction or indication of customer’s churn. If ML models are trained with such essential and crucial data, they can provide better prediction with high accuracy. This study experimented 16 ML models (single as well as ensemble) and the model performances were evaluated in two ways - 80:20 train-test split and 10-fold cross-validation. In 80:20 split, categorical boosting CatBoost classifier outperformed other models with 97.54% accuracy. However, Light Gradient Boosting Machine classifier LGBMC performed best in 10-fold cross-validation by achieving average accuracy of 96.39% with 1.08% standard deviation while CatBoost classifier was second best performing model. Thus, machine learning techniques can very effectively serve Telco industry as well as other customer-oriented competitive businesses. However, limitation of this study is that it experimented one dataset having 2850 distinct instances with 14 feature variables including target feature. Hence, further experiments with more and large datasets are suggested. Businesses including telecommunication industries can implement these ML models by deploying them in cloud or online platform. In this regard, an online database is required to maintain to automatically collect the from customers involving all the feature variables. If potential churn of any customer is predicted by the ML models, initiatives should be taken by providing improved customer care and personalized services to retain the customer. 
ISO-DBSCAN: An Unsupervised Approach to Anomaly Detection with NILM Data in Residential Buildings
No full textThe development of an efficient energy management solution is a challenging research area, particularly in addressing unusual appliance behavior, safety, and predictive maintenance in residential settings. Many studies have been limited to labeled datasets and the identification of hidden clusters in the datasets has been impractical. This study introduces ISODBSCAN, a hybrid unsupervised model that combines density-based spatial clustering of noise applications (DBSCAN) and isolation forest to detect anomalies in NILM data without labeled inputs. Using data from House 4 in the REDD dataset, we pre-process the time series energy readings and extract segments representing appliance behavior. DBSCAN is used to identify dense clusters of normal usage, while Isolation Forest further analyses these results to isolate outliers that deviate from expected patterns. The model achieves a silhouette score of 0.46, outperforming conventional methods like K-Means and Gaussian Mixture Models (GMM), which struggle with irregular cluster shapes and require predefined parameters. ISO-DBSCAN offers both robustness and interpretability, effectively isolating anomalies and mapping them back to specific time windows, thereby enhancing transparency and facilitating real-time monitoring. This approach is particularly beneficial in residential settings where energy behavior varies and labels are unavailable. The model’s performance is sensitive to DBSCAN parameters, such as minPts. Future work includes integrating automated hyperparameter tuning techniques, such as Bayesian optimization, expanding the approach to additional households and data sets, and improving real-time usability through dashboards and alert systems
AI-Enhanced Runtime Security for Kubernetes: Real-time Threat Detection with eBPF Telemetry and Ensemble Learning
No full textModern digital services are based on scalable, resilient, and adaptable Cloud-native ecosystems, particularly Kubernetes-based ones. Nonetheless, this fast usage has also widened the threat context and conventional security practices cannot be used in the fight against dynamic and distributed vulnerabilities. The incorporation of Artificial Intelligence (AI) and Machine Learning (ML) solutions into cloud-native security systems has emerged as an intriguing option for addressing these threats. The study is an in-depth discussion of AI-based approaches to improving access controls, anomaly detection, and intrusion prevention in Kubernetes environments. In order to demonstrate the potential of providing a defense mechanism that is adaptive, transparent, and resilient, it compares federated learning, reinforcement learning, graph neural networks, and explainable AI. The results of the current research indicate that the accuracy, efficiency, and scalability of real-time threat detection and policy enforcement using AI-based models can substantially surpass the of traditional ones. In addition, this paper provides insight into the role of self-healing architectures, forecast defense tools, and energy-saving artificial intelligence models in securing multi-cloud and hybrid applications. This work synthesizes the findings of the recent progress to give a coherent view of the development of AI-enhanced cloud-native security and presents both theoretical and practical conclusions regarding academia, industry, and security practitioners
Beyond Adoption: towards a Strategic Framework for Scaling Machine Learning in E-Commerce Cybersecurity
No full textThis study develops and validates a Machine Learning Scaling Framework (ML-SF) for e-commerce cybersecurity, addressing the critical gap between model adoption and sustainable operational scalability. Drawing on the Technology Organization Environment (TOE) framework and Dynamic Capabilities Theory (DCT), the study conceptualizes ML scalability as a socio-technical process that integrates governance, infrastructure, data readiness, continuous learning, and ethical oversight. Using three open cybersecurity datasets the CSE-CIC-IDS2018, the AI Incident Database, and the CIC-DDoS2019 the research employs rolling-window drift evaluation, Negative Binomial regression, and Interrupted Time Series analysis to examine post-adoption challenges and validate the proposed framework. Findings show that performance degradation occurs prior to detected model drift but recovers rapidly after retraining, confirming the importance of automated drift detection and retraining governance. Organizational analysis reveals that gaps in governance, data management, and MLOps maturity significantly increase incident severity, while the activation of ML-SF controls improves detection rates and reduces false positives over time. These results empirically demonstrate that scalable machine learning requires not just algorithmic innovation but institutionalized processes of monitoring, explainability, and adaptive resilience. The study contributes a theoretically grounded and empirically verified roadmap for scaling ML in cybersecurity environments. It offers actionable strategies for policymakers and enterprise leaders, including establishing AI governance boards, embedding always-on retraining pipelines, adopting federated learning for data privacy, and institutionalizing adversarial resilience as a compliance standard
Survey on Evolution of Java Web Technologies and Best Practices: from Servlets to Microservices
No full textThe evolution of Java web technologies has had a significant impact on the development of modern web and enterprise application development. This paper provides a thorough review of the history of Java since its establishment as a server-side set of technologies up to the present-day architectures based on microservices. It starts with the analysis of the initial innovations like Servlets, Java Server Pages (JSP), and Java Database Connectivity (JDBC) that defined the dynamic content generation, database integration, and simple web actions. It is then followed by the discussion of the emergence of enterprise frameworks, such as Struts, Spring, and Hibernate, which brought the concepts of modularity, scale, and better maintainability of applications with the separation of concerns. Further, the paper discusses the improvement of enterprise solutions by using Enterprise JavaBeans (EJB), the Java Persistence API (JPA), and by use of RESTful web services via JAX-RS, the roles that they play in providing strong data handling and the communication of services. Over the last few years, lightweight, cloud-native, and distributed systems have gained more and more traction, and approaches like Spring Boot, Quarkus, and Micronaut have helped to accelerate their development, containerization, and scaling. Furthermore, the use of new tools, such as Resilience4j, Open Telemetry, and Micro Profile, enhances the observability, resilience, and DevOps alignment of the enterprise world. There is a special section with a microservices architecture checklist that focuses on the configuration management, fault tolerance, service discovery, automation of testing, and data consistency. This study gives a detailed insight into how Java is still undergoing development as a flexible, stable and scalable platform to spearhead innovation in web engineering and enable scalable, secure and cloud-native enterprise applications by incorporating a historical and technical analysis
Neuro-Symbolic AI in Database Systems: A Systematic Review of Query Processing, Knowledge Graph Reasoning, Data Storage and Integration
No full textNeuro-symbolic AI is changing how database systems work by combining neural networks\u27 ability to recognize patterns in data with the logical reasoning of symbolic AI. This review looks at how these two approaches work together to improve database systems in areas like query processing, knowledge graph reasoning, and data representation. We also examine how studies from 2020 to 2025 measure performance, report improvements, and discuss limitations. Several studies show that neuro-symbolic methods can boost accuracy by 10–20% and cut costs by up to 80% compared to older methods. These systems are also better at handling unstructured data, dealing with incomplete information, and working without fixed schemas. Still, scalability is a significant challenge, with most studies noting problems as systems grow larger. Integrating these technologies can also be complex and often requires advanced engineering. In summary, while neuro-symbolic AI has great promise for database technology, more research and development are needed before it can be widely used in large organizations
Enhanced Zonal Stable Election Protocol for Wireless Sensor Network
No full textEnergy depletion in wireless sensor networks remains a concern for researchers in this emerging technology. However, several enhanced routing protocols are being designed to reduce the energy consumption in these networks and improve their lifetime. The Zonal Stable Election Protocol was carefully studied. It was observed that, no cluster head selection is done in zone 0 because of the nearest of some of the normal nodes to the Base station. This approach appeared effective for a short period of time, especially for the nodes that are really closer to the Base station. However, when some of the nodes in Zone 0 are a little farther away and the network operation is longer, the nodes are seen dying very quickly. Also, the existing scheme is observed to be transmitting captured data unnecessarily without considering the amount of data required by each node before transmission. In this research, a reactive heterogeneous routing scheme called the Enhanced Zonal Stable Election protocol (EZ-SEP) is proposed. The new scheme introduced clustering techniques in all the Zones. Cluster heads are selected based on their residual energy and relative distance from the base station. This is to ensure that the energetic heads that can transmit data to the Base station are selected. Furthermore, the new scheme introduced inter-cluster transmission and multi-hop communication to ensure effective energy distribution among the nodes while using hard and soft thresholds to reduce unnecessary data transmission to the Base station. Simulation was conducted to evaluate the performance of the new scheme using MATLAB R2018a, and the results showed that EZ- SEP outperformed Zonal SEP in terms of coverage, throughput approximately, 50%, stability period, residual energy and network lifetime
A Compact Reconfigurable Antenna-filter Co-design with Frequency Agility for 5G/6G Wireless Front-ends
No full textThe fast development of 5G and the next generation 6G networks requires small, spectrally adaptive and interference-free RF front-end modules. The proposed paper introduces a co-designed reconfigurable antenna-filter system that combines the radiating and filtering functions into a small-size unit of 2020 mm 2-a 43 percent smaller than the conventional designs. The system includes PIN-diode based tunability and shared-feed integration plan, to achieve multi-band operation of 2.8-4.2 GHz, and the return losses are less than -15 dB and the insertion losses in all bands are less than 2 dB. Radiation efficiencies exceeding 85 per cent, without any additional complexity, can be achieved in microstrip on high-permittivity materials with fractal-edged geometries. This is a full-wave co-simulation (ANSYS HFSS/CST) with the performance of the modeled and measured results being in good agreement. This solution, in contrast to previous solutions which required fixed or partially integrated designs, allows real time frequency agility and also has low power consumption and high integration density. It has been mentioned as a possible biomedical use, such as wearable sensors and implantable telemetry, but these are only mentioned in frequency compliance and size restriction; no clinical uses are mentioned. The study presents a verified design approach to the current state-of-art of compact and adaptive RF front ends to next-generation wireless and internet of things platforms. Lee et al., (2024)
ICT and Rural Banking: Evidence from Sri Lanka
No full textICT has transformed the banking industry by making online banking, mobile payments, and digital financial services more efficient and accessible. While it enhances user convenience and financial inclusion, considerable challenges remain, including cybersecurity risks, high costs, and low uptake in rural areas. This study aims to analyze the impact of ICT on rural Sri Lankan banking by concentrating on the transition to online banking, its implications for accessibility and efficiency, consumer difficulties with e-banking, and suggestions to enhance trust, usability, and financial inclusion. The researcher selected 100 bank customers and management officers from five major commercial banks in Anuradhapura district, North Central Province, Sri Lanka. To examine the usage, awareness and customer experience with e-banking and mobile banking, an anonymous open-ended questionnaire was distributed through personal contacts, social media and bank visits. Additional information about digital banking users was obtained from the participating banks. The acquired data were pre-processed to ensure completeness before analysis using SPSS. The questionnaire responses were categorized for thematic significance, allowing for qualitative and quantitative interpretations. Comparisons were made between banks to analyze the usage and trends of ICT in rural areas. According to the results, ICT initiatives have improved rural financial inclusion in Sri Lanka. However, despite widespread use among the youth, adoption of online banking, in particular, remains low due to low digital literacy, inadequate infrastructure, security concerns, and aversion to cash. ICT\u27s effects on Sri Lanka\u27s banking industry have mostly been studied in urban areas, with little attention paid to the rural experience. Despite the fact that districts have been examined, the results mostly represent urban viewpoints. To address this gap, this study looks at how ICT affects rural banking experiences using Anuradhapura as a case study
The Economic Impact of Internet Fraud among Students of Tertiary Institutions in the Upper East Region of Ghana
No full textThe prevalence of web-aided crimes in tertiary institutions, such as fraud, cyber harassment and hacking, is a growing concern. These criminal activities do not only tarnish the reputation of the institutions but also have far-reaching implications, impacting graduates\u27 job prospects, particularly those with backgrounds in Information Technology (IT). In light of this, a study was conducted using snowball and stratified sampling to explore the economic impact of internet fraud in the Upper East Region of Ghana. The results indicate factors such as age, income, or reasons for engaging in fraud increase, students’ likelihood of achieving a higher GPA decreases, demonstrating a negative academic impact. Although some students may have used financial gains from internet fraud to fund their education and living expenses, many are ultimately distracted from their academic goals, leading to poorer academic performance. Redirecting their Information Communication Technology (ICT) skills towards legitimate professions could significantly benefit the nation, particularly in facilitating successful business transactions. Students can also be exposed and trained to get involved in legitimate online opportunities and businesses through online data entry and surveys, software testing and advertising among others. Implementing training programs to guide these individuals in utilizing their skills for the country\u27s advancement is one potential solution. Further research can be conducted on students who engaged in fraud activities in senior high school and after tertiary education