Authors - Tushaar Yenduri, S S R Subramanya Hemant Konduri, Kalyan Netti Abstract - This study examines three inference strategies—serial, multi-core parallel (Joblib), and distributed memory (MPI)—to address scalability challenges in applying machine learning models to large tabular datasets. Using the HIGGS dataset as a representative benchmark of high-volume scientific data, a Random Forest classifier was trained and validated with performance metrics of 73.3% accuracy and 81.25% ROC-AUC. The results offer concrete guidance on selecting appropriate inference backends based on system architecture and dataset size, particularly for practitioners in high-performance computing and scientific machine learning domains.
Authors - Kevin De la Cruz, Alberto Taza, Edgar Ramos, Donovan Fuqua Abstract - The textile industry is facing growing pressure to transition toward production systems that are sustainable, efficient, and aligned with circular economy principles. This study develops a structured multi-criteria decision-making framework designed to integrate lean practices, circularity, and sustainability into strategic planning for textile manufacturing. The methodology applies the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) approach to evaluate and rank key operational practices considering their environmental, operational, and economic impacts. Results indicate that prioritizing lean initiatives with embedded circularity practices provides a dual advantage: improving short-term efficiency while simultaneously supporting broader sustainability goals. The proposed model offers decision-makers in the textile sector a practical and evidence-based tool to address challenges of competitiveness, environmental regulation, and sustainable value creation. Ultimately, this research contributes to advancing integrated planning approaches that strengthen the ongoing transition toward sustainable manufacturing systems.
Authors - Matana Charoenwongsa, Sangkama Sarawat, Laknara Chandraramya, Waralak Vongdoiwang Siricharoen, Thirachit Saenphon Abstract - The global media landscape is experiencing a profound shift, shaped by the rise of artificial intelligence (AI), growing cross-cultural content demands, and evolving patterns of work. This study investigates how active stakeholder participation can guide the development of an AI-integrated curriculum within global media education. Emphasis is placed on nurturing technological fluency, interdisciplinary perspectives, and ethically grounded innovation. Drawing on a curriculum design initiative from Thailand’s Global Media Industry and Innovation Master’s Programme, the research explores how AI-driven tools, media production technologies, and entrepreneurial approaches are brought together in educational practice. Insights were collected from over 50 stakeholders—including media entrepreneurs, policymakers, educators, and students—and examined through thematic content analysis. The research highlights three core priorities that shaped the curriculum design: strengthening technological fluency, deepening global media literacy, and encouraging social innovation through entrepreneurship. It proposes a collaborative model that addresses technological change while centring human agency, with practical implications for curriculum reform and the future of sustainable media ventures in an AI-influenced era.
Authors - Chethana N S, Anithadevi M D Abstract - We introduce a groundbreaking steganographic technique that merges non-linear cost evaluationmethods with binary triangular matrix structures to achieve superior data concealment in color images.Our approach tackles the inherent weaknesses of conventional linear cost models by implementing amultiplicative framework that more effectively maintains the statistical characteristics of images whileoptimizing embedding performance. The integration of structured binary matrices enables intelligentposition selection through organized randomization, preserving both visual fidelity and cryptographicsecurity. Our methodology incorporates sophisticated texture evaluation, self-adjusting parameter tuning, andmulti-channel analysis to deliver outstanding results across various image categories. Through rigoroustesting, we demonstrate remarkable improvements over current leading methods, achieving averagePSNR values of 52.34 dB and SSIM scores of 0.9847 across standardized test collections from multipleimage databases. The system exhibits substantially enhanced resilience against contemporarysteganalysis tools while supporting high embedding rates up to 0.8 bits per pixel with strengthenedsecurity assurances.
Authors - YURY MIKHEEV, URSULA PODOSENIN, SERGEY V. SYCHOV Abstract - This study investigates the reliability and validity of the 360-degree feedback method by focusing on three primary sources of bias: respondent roles, interpersonal relationships, and individual evaluative styles. Drawing on data from over 15,000 respondents across multiple organizations, it applies a Python-based statistical algorithm to quantify the impact of these biases on competency ratings. Findings reveal that such factors can account for up to half of the variance in assessments. Mitigation approaches include evaluator anonymity, structured training, and correction coefficients. With these measures in place, 360-degree feedback continues to serve as an effective tool for fair and accurate performance appraisal, supporting personnel decisions and sustainable talent management.
Authors - Zarif Wasif Bhuiyan, Zaed Bin Monir Atib, Mahady Hasan, Farhad Alam, Md. Tarek Habib Abstract - Early and accurate identification of common skin conditions is essential for appropriate therapy and better clinical outcomes. However, diagnostic expertise is often limited in resource-constrained healthcare or remote settings. We present a robust deep learning approach based on a custom convolutional neural network (CNN) to automatically classify five prevalent skin diseases (acne, contact dermatitis, nail fungus, scabies, and urticaria) from clinical images. A curated dataset of 8,900 clinical images (1,780 per class) was compiled, with standardized resizing, normalization, and extensive augmentation applied to simulate real-world variability and enhance generalization. The CNN architecture consists of five sequential convolutional blocks with progressively increasing filters to extract high-level features, accompanied by dropout layers to mitigate overfitting, and includes a fully connected layer prior to the SoftMax output. On evaluation, the model achieved an overall classification accuracy of 99%, demonstrating robust performance and high diagnostic precision across all categories. It notably outperformed conventional machine learning classifiers as well as state-of-the-art pretrained deep networks on the same task. This exceptional performance, combined with strong generalization capabilities, underscores the model’s reliability and suitability for deployment in real-world clinical practice and tele dermatology settings. In such scenarios, rapid and accurate screening for multiple skin conditions can significantly improve patient care and enable earlier interventions.
Authors - Amruta Khanaj, Sourabh Khanaj, Smita S. Sangewar Abstract - This research investigates the security implications of reusing code snippets of Stack Overflow in open-source software projects. We present a novel framework that proactively identifies and mitigates security risks associated with outdated Stack Overflow code snippets. Our methodology combines static code analysis techniques, continuous monitoring of Stack Overflow post changes, and an automated developer notification system. We analyzed 1.5 million code snippets from Stack Overflow approximately across 5000 popular open-source GitHub projects, focusing on programming codes. Our observations reveal that there are a significant number of reused snippets becoming outdated over time, with developers often missing critical security updates made to the original Stack Overflow posts. We showcase the effectiveness of our framework in detecting potential vulnerabilities and alerting developers to security relevant updates in reused code. The research emphasizes the need for continuous monitoring of community-sourced code and suggests solutions to ensure its secure integration in software development practices. Our work contributes to the broader understanding of software supply chain security and the evolving nature of code reuse in the open-source ecosystem.
Authors - Mst.Sadia Siddika, Sherin Sultana, Md. Tarek Habib, Sabrina Alam, Md. Mahbub Alam, Mahady Hasan Abstract - Agricultural products are graded based on their quality and adherence to standards. Managing the grading, packaging, and transportation of products, especially perishable items like food, is crucial for maintaining freshness and safety. Packaged goods are transferred to storage facilities and prepared for transportation. To maintain product freshness, a well-organized and traceable supply chain is essential. This study demonstrates how technologies such as the Internet of Things (IoT) can facilitate an agricultural supply chain system that uses maps, computer vision, smart packaging, and smart barcodes. These technologies help track products in real-time, ensure data security via databases, and streamline supply chain operations through web-based platforms. Using an app-based system, the solution highlights traceability, automation, and user-friendly interfaces to optimize efficiency from farm to market. The app also provides automated alerts and real-time inventory updates for both customers and sellers. Additionally, the system should integrate IoT-based temperature and humidity monitoring to maintain the quality of perishable goods. Overall, this study suggests that combining these advanced technologies into a unified system can address challenges in traditional methods, such as delays and product quality loss.
Authors - Kostaryev D.B., Tevyashev A.D., Sizova N.D., Tkachenko V.P. Abstract - An intelligent information system for proactive management of a printing enterprise based on artificial intelligence is proposed. Approaches to proactive enterprise management grounded in Industry 4.0/5.0 are substantiated, including the use of relevant scientific methods and mathematical models. The architecture of the information system, based on mathematical models and scientific principles, has been developed. The results of evaluating the effective-ness of implementing and using this intelligent information system at a printing enterprise are presented.
Authors - Zarif Wasif Bhuiyan, MD. Humayun Kabir, Paramita Saha, Md Mahbub Alam, Mahady Hasan Abstract - Urban–rural evidence in Bangladesh is often fragmented or short-lived, limiting its value for tracking change. This study leverages the Life in the Field Experience at Independent University, Bangladesh to build a pre-COVID baseline from eight annual waves of student-collected surveys spanning 2013–2020 across six districts: Jessore, Bogra, Mymensingh, Sylhet, Barishal, and Dinajpur. Heterogeneous spreadsheets are harmonized into standardized sector indices for education, medications and health service use, employment, marital history, acute morbidity, and chronic morbidity. A transparent pipeline handles header inconsistencies, variable shifts, and missingness, producing comparable annual series. ARIMA models are estimated for Jessore to generate five-year projections and are interpreted alongside cross-district benchmarking using z-scores. Results show steady gains in education that level off, an employment dip followed by recovery, rising medication use, low and stable acute morbidity, a gradual increase in chronic morbidity, and largely stable marital patterns. Forecasts are framed as extensions of pre-COVID dynamics with explicit uncertainty. The pipeline provides a reusable template for short annual series and supports policy diagnosis at district level. Future work will extend forecasting to all districts, incorporate structural-break tests and hierarchical models, and enrich indices with external covariates and administrative validation.
