Asian Journal of Research in Computer Science
Not a member yet
792 research outputs found
Sort by
Application of Stacked Ensemble Techniques for Classifying Recurrent Head and Neck Squamous Cell Carcinoma Prognosis
No full textA feature selection technique should, in theory, be able to reliably extract pertinent features, identify non-linear feature interactions, scale linearly with the number of features and dimensions, and permit the integration of known sparsity structure. Identifying a machine learning algorithm that performs best given varied distributions may be quite challenging because not all machine learning algorithms are equally created, even though many of them suit very well for a given task. The heterogeneous ensemble feature selection (HETR-EFS) technique learns to combine the feature subsets provided by base feature selectors in an ensemble. Similarly, the stacked ensemble (SE) technique learns how to best combine base classifier models to form a strong model. As a prognostic model for classifying Head and Neck Squamous Cell Carcinoma {HNSCC} recurrence patterns, this study sought to identify the combination of SE classification model and EFS technique that fit optimally when the same ML classifiers for EFS and SE learning are used. Four SE classification models; in which first one used two base classifiers: gradient boosting machine (GBM) and distributed random forest (DRF); second one used three base classifiers: GBM, DRF, and deep neural network (DNN); third one used four base classifiers: GBM, DRF, DNN, and generalized linear model (GLM); and fourth one used five base classifiers: GBM, DRF, DNN, GLM, and Naïve bayes (NB), were developed based on various EFS techniques, using GBM meta-classifier in each case. The results showed that implementing SE technique consisting of five base classifiers on heterogeneous ensemble feature (HETR-EF) subset achieved better performance than achieved on other EF subsets and implementing this SE technique on HETR-EFs achieved better performance compared to other SE techniques implemented on HETR-EFs and other feature subsets used. Thus, learning SE technique having five base classifiers on HETR-EFs is clinically appropriate as a prognostic model for classifying and predicting HNSCC patients’ recurrence data
Advanced Battery Management: Forecasting Health, State of Charge & Maintenance Needs Using AI & ML Models (LSTM, Gradient Boosting, SVR, Random Forest)
No full textThe rapid expansion of renewable energy sources and the widespread adoption of electric vehicles underscore the critical demand for efficient energy storage systems. This conference paper explores cutting-edge predictive models tailored for forecasting battery health, State of Charge (SOC), and anticipating maintenance requirements. Employing advanced machine learning [1,2] techniques, innovative feature engineering, and rigorous evaluation metrics, the study achieves robust performance in predicting key aspects of battery behavior. Key methodologies include Stacked LSTM networks, Random Forests, Gradient Boosting, and SVR. Alongside advanced time series analysis methods like ARIMA and SARIMA.
The results demonstrate significant advancements in SOC prediction accuracy and provide valuable insights into overall battery health assessment. The models effectively identify potential maintenance needs, representing a substantial integration of machine learning [1,2] and time series analysis for enhanced battery management. These developments hold profound implications for energy storage and management, benefiting industries reliant on energy-intensive processes such as manufacturing, IT Infrastructure & Data Centers etc. They optimize energy usage, reduce costs, and enhance service efficiency and uptime in the retail sector, particularly for electric vehicle servicing.
This research underscores the transformative impact of advanced predictive modeling on energy storage and management, supporting sustainable practices and fostering innovation across industries
Strengthening Cybersecurity Measures for the Defense of Critical Infrastructure in the United States
No full textThis study addresses critical cybersecurity vulnerabilities within U.S. infrastructure sectors, particularly energy, water, and healthcare, where high-severity vulnerabilities and ransomware continue to pose significant risks. The study applied a multi-method analytical approach comprising logistic regression, K-means clustering, Interrupted Time Series (ITS), Difference-in-Differences (DiD), and Kaplan-Meier survival analysis to identify, prioritize, and evaluate vulnerabilities across these essential sectors to provide insights into sector-specific risks and framework effectiveness. Logistic regression models were specifically used to quantify the likelihood of incidents by examining vulnerability attributes. In contrast, K-means clustering was used to access insights into patterns of shared vulnerabilities unique to each sector. The ITS and DiD analyses were also used to measure the National Cybersecurity Strategy’s effect, showing a 3.7% reduction in incidents post-intervention, particularly within the healthcare sector. Furthermore, Kaplan-Meier survival analysis was used to assess how long systems withstand attacks, highlighting that ransomware has the most immediate and costly impact, with average recovery costs reaching $540,000 per incident. These findings aver the need for proactive cybersecurity defences across critical infrastructure, where the potential for disruption directly impacts public safety and economic stability. To strengthen resilience, the study recommends tailored, sector-specific cybersecurity frameworks, the prioritization of high-risk vulnerabilities, a reinforced zero-trust architecture, and expanded public-private collaboration for real-time threat intelligence sharing, as adopting these strategies in the U.S. can contribute to developing a more adaptive cybersecurity infrastructure capable of countering evolving threats
Chaotic Dynamics of a Two-Neuron Memristive Hopfield Neural Network under Electromagnetic Radiation
No full textElectromagnetic radiation (EMR) is a recognized approach for investigating the behavior of the nervous system. This research explores a Hopfield Neural Network (HNN) with two neurons, examining the interplay between synapses and hyperbolic memristors, along with the effects of EMR. By modifying the interference among the networks and adjusting synaptic weights, we can control neuronal capabilities. The study simulates synaptic interference between the two neurons, incorporating parameters of weight and memory, and analyzes how EMR influences chaotic dynamics, complex behavior, transient disturbances, phase portraits, chaotic phenomena, and branching diagrams within these neural networks. This paper investigates how electromagnetic radiation (EMR) influences chaotic dynamics in a two-neuron-based memristive Hopfield neural network (HNN) with synaptic crosstalk. The dynamic behavior of the HNN can be regulated by altering the EMR input to the affected neuron. The proposed model has been simulated using PSpice. The findings demonstrate that external stimuli, represented by EMR, can both enhance complex dynamic behaviors and suppress chaotic patterns by adjusting parameters. Finally, circuit experiments using PSpice confirm the feasibility of the theoretical model, contributing to the control of chaotic phenomena
Analyzing Different Architectures of Convolutional Neural Networks for Tomato Grading System
No full textThe tomato is a very popular and commonly eaten fruit. Its quality, which affects how people see it, depends a lot on how it looks. Convolutional neural networks, which are advanced computer programs, are great at using deep learning to sort and classify fruits, grains, and vegetables in farming. Right now, there are two ways to classify tomatoes: by eye or by analyzing images. The first method, which involves checking tomatoes by hand, is more accurate but takes longer and costs more. The second method, which uses images, is faster and cheaper but not as precise. In this study, we use a deep learning approach to classify tomato quality, specifically using convolutional neural networks (CNNs). We compared two popular CNN models, ResNet-50 and AlexNet, and tested how well these models automatically find important features in the tomatoes. The success percentage of our suggested strategy in experiments was 99.1%. Our proposed method outperforms existing image-processed tomato quality rating systems on all five of the commonly used evaluation criteria, including accuracy, precision, recall, specificity, and F-score
Enhanced Text Recognition in Images Using Tesseract OCR within the Laravel Framework
No full textThis research explores the integration of Tesseract OCR (Optical Character Recognition) within the Laravel framework to enhance text recognition capabilities in images. Tesseract OCR, an open-source OCR engine, is renowned for its accuracy and efficiency in converting various image formats into editable and searchable text. However, leveraging its full potential within a robust web application framework presents unique challenges and opportunities. This implementation focuses on creating a seamless, user-friendly application that processes images uploaded by users and accurately extracts text content. The Laravel framework, known for its elegant syntax and extensive ecosystem, serves as the backbone of our application, ensuring scalability, security, and maintainability. Key features of our system include image preprocessing techniques to improve OCR accuracy, handling different languages and fonts, and providing real-time feedback to users. This research delves into the specifics of integrating Tesseract with Laravel, detailing the process of setting up the environment, managing dependencies, and optimizing the OCR process for web applications. This project also addresses common issues such as noisy images, varied text orientations, and low-resolution graphics, employing advanced preprocessing methods like binarization, deskewing, and noise reduction. Performance benchmarks demonstrate significant improvements in text recognition accuracy and processing speed. Additionally, a comparative analysis with other OCR solutions to highlight the advantages of the used approach is provided. The application’s effectiveness is further validated through diverse use cases, ranging from digitizing historical documents to extracting text from natural scene images. Ultimately, this research contributes to the field by presenting a comprehensive, practical implementation of enhanced text recognition in images using Tesseract OCR within the Laravel framework. The findings suggest that with proper integration and preprocessing, Tesseract’s capabilities can be significantly amplified making it a powerful tool for various text recognition applications in web development
How can Blockchain be Integrated into Autonomous Systems to Ensure Data Integrity and Trustworthiness, and What are the Potential Pitfalls in Decentralized Autonomous System Operations?
No full textReal-time applications always have security issues present in their ecosystem which needs to be improved. Blockchain technology can be considered as a solution that can ensure network security. The integration of blockchain technology into autonomous systems has now become an important candidate to resolve network security issues in devices that work on p2p communications. As humans become more reliant on autonomous vehicles, the importance of finding a solution to the problems linked with data security needs to be addressed to comply with international security laws. The paper aims to find the importance of blockchain technology and how it can be integrated into different classes of autonomous vehicles. In the research, an analysis of existing papers will be included along with a proposed design that can be deployed to solve the problems associated with data security using blockchain technology. Besides the improvements offered by blockchain technology, all the additional challenges associated with designing a new system and deploying this technology will be reviewed. The limitations of the works along with the future research opportunities that might open up will be presented in this paper
ReT-ELBa: A Novel Algorithm for Efficient Load Balancing in Cloud Computing
No full textThis paper presents a novel load-balancing algorithm for balancing load on the cloud environment which we call Response Time Efficient Load Balancer (ReT-ELBa). Load balancing involves a dynamic and equal distribution of workloads among processors or Virtual Machines to achieve better resource utilisation. The study gives an insight into the design, implementation and evaluation of an enhanced load-balancing algorithm that allocates tasks across Virtual Machines on the cloud, which minimises Response Time and increases resource utilisation. The ReT-ELBa distributes tasks based on their sizes, the requirements for their executions, as well as the state of the Virtual Machines. The Cloud Analyst simulation tool was employed to simulate or evaluate various cases for the ReT-ELBa in comparison with the Throttled and Round Robin algorithms. The study revealed that the ReT-ELBa outperformed the two algorithms in relation to Response Time. The ReT-ELBa also outperformed the two algorithms in relation to Data Centre Processing Time for all simulated cases, except one (Case 3), which recorded a small difference of 0.07ms. Even the Response Time was the Primary focus of this research, the Data Centre Processing Time of the algorithm was also analysed to present a more elaborate representation of the performances of the algorithms
Influence of Variation in Tie Reinforcement Diameter on the Ductility of Reinforced Concrete Columns
No full textReinforced concrete columns are crucial structural elements in ensuring the strength and stability of buildings. Column ductility, which is the ability to absorb energy and undergo deformation before failure, is a primary concern in structural engineering, especially in extreme external loading situations such as earthquakes. This study aims to evaluate the influence of variation in Tie reinforcement diameter on the ductility of reinforced concrete columns using Xtract software. The research method involves creating column structure models with specified dimensions and specifications, followed by the gradual application of axial loads to each model. Three models were created with varying Tie reinforcement diameters, namely 10 mm, 12 mm, and 14 mm. Structural analysis was conducted to examine the structure\u27s response to the applied axial and moment loads, including evaluation of stresses, deformations, and column capacities. The analysis results show differences in ductility levels among the models. The model with a 10 mm Tie reinforcement diameter achieves higher ductility levels at low axial loads but fails to meet the requirements at higher axial loads according to the SNI 1726:2019 standard. However, models with Tie reinforcement diameters of 12 mm and 14 mm also exhibit a similar pattern, with good ductility at low axial loads but failing to meet the requirements at higher axial loads. In conclusion, variations in Tie reinforcement diameter affect the ductility of reinforced concrete columns. To ensure full ductility under various axial load conditions, adjustments to the design or specifications of the reinforced concrete column structure are necessary. This research contributes to understanding the factors influencing the ductility of reinforced concrete columns and can serve as a basis for the development of more effective design methods in the future
Digital Collaborative Tools, Strategic Communication, and Social Capital: Unveiling the Impact of Digital Transformation on Organizational Dynamics
No full textThe rapid advancement of digitalization has significantly impacted how organizations operate, with collaborative tools like Asana, Trello, and Slack becoming integral to fostering internal communication, collaboration, and social capital. Despite their widespread adoption, the comprehensive impact of these tools on organizational culture, employee engagement, and social capital remains underexplored. This study aims to bridge this gap by examining the potential of digital collaborative tools to enhance organizational social capital and exploring the role of strategic communication and digital inclusivity in this process. Utilizing a quantitative research approach, data were collected from 557 professionals through a survey distributed via LinkedIn. The questionnaire, designed with Likert scale closed-ended questions, focused on the utilization of collaborative tools and their perceived impact on communication, collaboration, and social capital within organizations. Multiple regression analysis was employed to test four hypotheses related to the effects of digital tool usage on organizational dynamics. The findings reveal that digital tools significantly enhance communication and collaboration within organizations, with the frequency and purpose of tool usage being key factors. Moreover, digital tool usage for networking and the quality of interactions facilitated by these tools were found to positively impact the development of social capital. However, challenges in integrating digital tools were shown to negatively affect this development, though technical support and training could mitigate these effects. Interestingly, the introduction of a \u27social score\u27 feature within digital tools was perceived positively, indicating its potential to further engage users and enrich organizational social capital. This study underscores the importance of adopting a strategic approach to implementing digital tools, emphasizing their role beyond operational efficiency to include the enhancement of social interactions and organizational culture. Recommendations include fostering digital tool usage for networking, providing adequate technical support and training, and exploring the integration of \u27social score\u27 features to incentivize engagement and collaboration