Asian Journal of Research in Computer Science
Not a member yet
792 research outputs found
Sort by
Overview of Algorithms for Image Recognition
No full textThe significance of image recognition technology is highlighted by its wide applications in fields such as security, medical image analysis, and data analysis. Its growing popularity reflects advancements in research. Traditional machine learning methods have markedly improved feature extraction, while deep learning techniques have advanced significantly due to the application of various neural networks. This paper reviews algorithms and systems for image recognition, covering both traditional and deep learning methods. It provides extensive descriptions of classification and object detection techniques involving feature extraction, convolutional neural network designs, and neuron activation functions. The focus extends to traditional algorithms like k-nearest neighbor, support vector machine, Naive Bayes, and parallel cascade selection. Additionally, it explores various deep learning approaches for image interpretation, detailing different convolutional network dimensions and neuron model constructions. The paper concludes by illustrating algorithms with application examples and clarifying the differences between traditional methods and deep learning
Performance Optimization Techniques for Microservice Architectures in High-Load Scenarios
No full textThe article addresses existing methods for enhancing the performance of microservice architectures under high-load conditions, where stability and scalability are required to adapt to changing demands. The objective of the study is to systematize existing optimization methods. The methodological framework includes data analysis, a comparison of various approaches such as containerization, auto-scaling, and the use of frameworks for asynchronous request processing. The research was conducted based on an analysis of publicly available articles, providing a comprehensive examination of the topic. The analyzed studies demonstrate that implementing hybrid solutions incorporating machine learning for load forecasting and dynamic infrastructure configuration significantly improves performance. Additionally, the studies address the management of service states and interactions, which is critical for maintaining data integrity under high loads. The information presented in the article will be valuable for system architects, DevOps engineers, and cloud computing specialists working with resource-intensive services. These solutions enable the creation of scalable, reliable infrastructures capable of efficiently handling large volumes of real-time data. The conclusion confirms the necessity of a comprehensive approach to optimizing microservice systems, focusing on dynamic adaptation and the integration of new technologies
Dynamic Scaling and Performance Optimization for Microservices using Kubernetes
No full textThis study evaluates Kubernetes\u27 role in managing microservices under high-load conditions, emphasizing the efficiency of Horizontal Pod Autoscaler (HPA), Cluster Autoscaler, and security mechanisms. The research demonstrates how Kubernetes enhances scalability, reduces failure risks, and ensures stable performance. Experimental results validate its effectiveness in optimizing CPU load and response time for fluctuating workloads. The problems faced by developers when implementing this software were also considered, namely: security settings, optimization of auto-scaling, and configuration scheme of network interaction between components.
The experimental results presented in Tables 1–3 confirm the effectiveness of automatic microservices scaling in Kubernetes. Under loads of 10,000 and 100,000 connections, the average CPU load without scaling reached 606.34–555 ms, whereas with Horizontal Pod Autoscaler (HPA) enabled, this metric was reduced to 219–293 ms. Similarly, server response time in scenarios 1 and 2 decreased by more than half (from 43 to 12 ms and from 58 to 32 ms, respectively). These findings demonstrate that HPA and Cluster Autoscaler mechanisms, designed for dynamically adjusting the number of nodes based on the current load, optimize computational resources and enhance system responsiveness even under increasing traffic.
The article targets developers and software architects optimizing microservice applications. In conclusion, recommendations are provided on leveraging Kubernetes to build a flexible, fault-tolerant microservice architecture capable of handling high loads.
The article is aimed at developers and architects of software systems that optimize microservice applications. In conclusion, recommendations are given on using Kubernetes to create a flexible, fault-tolerant microservice structure ready for high loads
Advances in Skin Cancer Detection Using Machine Learning: Current Methods and Future Directions
No full textWith increasing incidence rates, high mortality risks, and substantial cost burdens, skin cancer is a serious global health concern. In order to improve patient outcomes, early and accurate detection is essential. Due to their heavy reliance on clinical knowledge, traditional diagnostic techniques are prone to subjectivity. In order to overcome these obstacles, automated skin cancer diagnosis has been using machine learning (ML) and deep learning (DL) models more and more. Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and ensemble learning models are among the ML and DL models that are methodically assessed and contrasted in this study for their ability to classify skin lesions. We examine how classification performance is affected by preprocessing methods, optimization tactics, and dataset selection. More specifically, this study makes use of publically accessible benchmark datasets including PH2, ISIC, and HAM10000 to guarantee a thorough assessment of model effectiveness. Our results show the benefits and drawbacks of various approaches, offering guidance for creating AI-driven diagnostic tools that are more precise, understandable, and useful for actual clinical settings
A Comprehensive Review of the Ford-fulkerson Algorithm for Network Flow Problems
No full textThe Ford-Fulkerson method is one of the basic algorithms for computing the maximum flow in a flow network. This identifies more frequent traversed paths as a means of maximizing the flow between a source node and a downstream node in a directed network. Therefore, this paper will present a literature review, the principles of the algorithm, its mathematical foundation, applications and improvements made to the algorithm, such as the use of breadth-first search (BFS) as in the Edmonds-Karp algorithm, parallel computing techniques and predictive modeling to enhance efficiency. However, the algorithm is not without its issues, such as the effect of dense networks, loss of correctness in capacities, and static structure of networks. Furthermore, it is compared with other algorithms like Dinic and Edmonds-Karp to understand their relative benefits and drawbacks. We conclude that the Ford-Fulkerson algorithm is still a basis method with broad application in many fields, including traffic network, logistics, and communication networks, but its performance can be greatly improvedas a result of the advances in modern computation approaches including parallel computing and adaptive dynamic path-propagation. Moreover, comparative studies showcase that alternative algorithms such as Dinic and Edmonds-Karp provide distinct advantages in specific settings, such as in dense and dynamic networks, underscoring that depending on the application the most effective algorithm may vary
The Impact of Cybersecurity Governance on National Security by Strengthening Critical Infrastructure through IT Auditing and Risk Management
No full textCybersecurity governance is increasingly recognized as a cornerstone of national security, especially in protecting critical infrastructure sectors such as energy, healthcare, finance, telecommunications, and transportation. This study investigates how governance frameworks, IT auditing, and risk management practices collectively reduce cyber threats and enhance the resilience of essential services. Using data from CISA, GAO audit reports, Verizon DBIR, and the World Economic Forum, the study assesses governance effectiveness through a combination of statistical techniques, including regression analysis, survival modeling, and data reduction methods. Findings reveal that organizations adopting both the NIST Cybersecurity Framework and ISO 27001 report a 75.8% reduction in cyber exploits, while IT audits lead to a 38–45% decrease in identified vulnerabilities. Additionally, proactive risk management strategies significantly delay the occurrence of cyber incidents, extending the average time to breach by over 260 days. These results underscore the critical importance of structured cybersecurity governance in minimizing threats and ensuring the continuity of national infrastructure. However, the study also highlights several implementation challenges, including regulatory inconsistencies, budget constraints, and a shortage of skilled cybersecurity professionals. These obstacles vary by region and sector, with under-resourced public institutions and developing economies facing the most significant barriers to effective governance. Recommendations include regulatory harmonization, mandating regular cybersecurity audits, and increasing investments in cybersecurity training and threat intelligence particularly in regions with fragmented oversight. The study offers valuable guidance for policymakers, regulators, and industry leaders seeking to strengthen national resilience in an evolving cyber threat landscape
Using Big Data and Machine Learning to Predict Household Appliance Failures: A New Approach to Preventive Maintenance
No full textThis study explores the application of big data processing methods and machine learning algorithms for predicting household appliance failures, shifting from traditional reactive maintenance models to proactive preventive repair systems. The study is based on an analysis of historical data, error logs, and sensor readings, enabling the identification of hidden patterns that indicate potential malfunctions. The novelty of this study lies in the adaptation and comprehensive application of modern data analysis methods to the operational specifics of household appliances, as well as an assessment of the economic efficiency of such solutions. The applied methodology includes stages of data collection and preprocessing, feature engineering, machine learning algorithm implementation, and comparative economic analysis. The results demonstrate the potential of predictive maintenance in reducing downtime, optimizing repair costs, and improving service quality. The findings of this study are valuable for data analysts, household appliance engineers, predictive model developers, and companies engaged in appliance servicing and manufacturing, seeking to enhance preventive maintenance efficiency using advanced machine learning and big data analytics methods. This study aims to shift from reactive maintenance to proactive preventive maintenance, addressing real-world challenges faced by appliance manufacturers and service providers
Biometric Authentication in Android: Enhancing Security with AI-Powered Solutions
No full textAims: This study aims to analyze biometric authentication methods on the Android platform, focusing on enhancing security through ready-to-use AI solutions. The research evaluates existing biometric authentication techniques, their vulnerabilities, and the application of AI-driven approaches to mitigate security risks.
Study Design: This is a review and analytical study that examines current biometric authentication mechanisms, AI-based enhancements, and their impact on security and accuracy.
Place and Duration of Study: The study is based on literature review and practical analysis of AI-enhanced biometric authentication methods applied in real-world Android applications.
Methodology: The research explores the evolution of biometric authentication in Android, emphasizing the use of AI-driven tools such as ML Kit for Face Detection, TensorFlow Lite, and OpenCV. The study assesses the effectiveness of these technologies in improving recognition accuracy, reducing false acceptance and rejection rates, and addressing security threats such as spoofing attacks. Performance metrics, including False Acceptance Rate (FAR), False Rejection Rate (FRR), and processing time, were used to evaluate AI-enhanced solutions.
Results: The findings indicate that AI-based enhancements significantly reduce the FAR by 15–20%, improving the overall reliability of biometric authentication. Machine learning models and image preprocessing techniques help adapt authentication to varying conditions, such as poor lighting and occlusions. However, AI integration introduces increased computational overhead, slightly extending processing time from 500ms to 700–800ms. Hardware-backed security measures mitigate risks associated with biometric data storage and manipulation.
Conclusion: AI-driven biometric authentication methods substantially improve security and accuracy on Android devices, addressing key vulnerabilities in traditional biometric techniques. Despite minor processing time increases, the trade-off is justified by enhanced protection against spoofing attacks and improved adaptability to environmental conditions. Future research should focus on optimizing AI models for mobile efficiency and developing multi-factor authentication approaches to further enhance security
Cloud-Based Web Applications for Enterprise Systems: A Review of AI and Marketing Innovations
No full textThrough the integration of big data analytics, advanced artificial intelligence, and scalable cloud architectures, this paper explores how cloud-based web applications are revolutionizing enterprise systems. It draws attention to how dynamic cloud environments, which improve operational efficiency, facilitate real-time decision-making, and support targeted marketing campaigns, are replacing conventional on-premise systems. The report explains how tailored analytics and automation driven by AI enhance service delivery while also promoting innovation and a competitive advantage. Additionally, it tackles important issues that need to be resolved in order to fully utilize these digital technologies, like cybersecurity, system integration, and governance. All things considered, the article provides a thorough analysis of digital transformation in contemporary businesses, highlighting the strategic collaboration of cloud computing, edge technologies, and artificial intelligence as a driver of long-term expansion and improved corporate performance
Enhancing Customer Churn Prediction in Telecommunications through Deep Learning: A Comprehensive Review
No full textCustomer churn remains a critical challenge for the telecommunication industry, directly impacting revenue and customer retention strategies. Traditional churn prediction models based on statistical and machine learning techniques have shown limited adaptability in capturing complex behavioural patterns. Deep learning (DL) methods, particularly recurrent neural networks (RNN), convolutional neural networks (CNN), and transformer-based architectures, have emerged as powerful tools for modelling customer churn by leveraging vast and dynamic datasets. By enabling precise identification of at-risk customers and personalized intervention strategies, DL-driven approaches not only enhance retention rates but also unlock targeted revenue-generating opportunities through tailored service upgrades and dynamic pricing models. This paper presents a comprehensive review of deep learning-based churn prediction mechanisms in the telecommunication sector, comparing their architectures, feature engineering strategies, and performance metrics. Highlights of recent advances in DL techniques, including attention mechanisms and explainable AI, were presented and their implications for improving customer retention strategies were discussed. Finally, key research challenges and future directions—such as developing DL models with simpler explainability frameworks, advancing techniques for class imbalance mitigation, and designing adaptive architectures for real-time, resource-efficient inference—were highlighted to bridge the gap between theoretical innovation and scalable deployment in operational environments