Volume 14 | Issue 4

 Abstract

Teacher engagement and turnover intention in higher education have emerged as critical challenges affecting institutional stability, academic quality, and student outcomes. While prior research has extensively examined teacher retention, fewer studies have explored the interconnectedness of various factors influencing teacher engagement, job satisfaction, and turnover intentions. This review paper synthesizes existing literature to investigate the interrelationships among job stress, organizational commitment, person-environment fit (PEF), and organizational citizenship behavior (OCB). The findings highlight that job stress negatively impacts teacher well-being and engagement, while strong organizational commitment enhances teacher retention. OCB serves as both a facilitator and a potential stressor, enhancing institutional loyalty while also leading to role overload. The intricate relationships among these variables highlight the necessity for higher education administrators to implement integrated retention strategies that balance workload, foster institutional commitment, and enhance teacher well-being. This study emphasizes the importance of a comprehensive approach in addressing reduced teacher engagement by considering the dynamic interactions between these variables. The review provides educational administrators with actionable strategies to improve teacher engagement, mitigate job stress, and create supportive academic environments that enhance institutional effectiveness. Future research should explore longitudinal and cross institutional studies to develop a more nuanced understanding of teacher engagement dynamics in higher education.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Teacher engagement, organizational commitment, job stress, person-environment fit, organizational citizenship behavior, turnover intentions

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Rural electrification remains a challenge that faces significant economic, geographical, and infrastructural hurdles, making the application of centralized power systems less effective. In this regard, the remoteness of the areas, low population density, high transmission costs, and unstable extension of power grids to rural areas make traditional electrification approaches inefficient and economically unviable. As a result, a significant proportion of the rural population remains unserved or unconnected to the power grid. In this regard, microgrids emerge as a feasible and sustainable decentralized approach that allows for the local production and distribution of electricity according to rural energy demand. This paper extends the application of microgrids not only as a tool for enhancing energy access but also as a viable business model that can sustain itself without external subsidies in the long term. The paper takes a business model approach to investigate the feasibility of financial sustainability through the application of renewable energy-based microgrids for the provision of reliable and quality electricity. One of the main areas of research in the study is the application of hybrid renewable energy harvesting, where different forms of renewable energy, such as solar, wind, and biomass, are combined to improve the reliability and resilience of the energy system. Hybrid systems provide a continuous supply of power by optimizing energy production based on environmental conditions, which is essential in rural areas. The paper also highlights the importance of electricity modification using capacitor-transformer configurations to improve the power quality, voltage regulation, and energy efficiency of rural microgrids. Electricity modification reduces transmission losses, improves power factor, and provides a stable voltage supply, which is essential to prevent damage to equipment and improve system performance. Another important aspect of the study is the storage of excess energy, which allows the use of surplus renewable energy produced during off-peak hours. Energy storage systems, such as batteries or other energy storage solutions, improve load balancing, supply reliability, and peak demand management, which is essential to improve the technical and economic viability of the system. In order to further improve the stability of revenues, the report recommends the inclusion of productive uses of energy. By coordinating electricity supply with rural economic activities such as agricultural processing, irrigation and water management, cold storage, cottage industries, and rural enterprises, microgrids provide stable and predictable demand. This productive use of energy helps to improve load factors and also fosters economic development in rural areas, thus creating a positive feedback cycle between energy access and rural development. The strategy adopted focuses on effective community engagement and participation to improve social acceptance, efficiency, and sustainability. Community participation in the operation, management, and governance of the system helps to improve accountability, capacity building, and risk reduction. Additionally, a scalable system design allows microgrids to scale up incrementally as demand increases, thus improving adaptability to changing energy demands. Environmental sustainability is also considered through the use of clean and renewable energy resources, which significantly reduces greenhouse gas emissions while improving energy security and climate change resilience. The report also highlights the need for enabling policy frameworks, entrepreneurship, innovative ownership structures, and access to low-cost capital as key facilitators for the scaling up of rural microgrids in different socio-economic settings.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Rural Electrification, Decentralized Energy Systems, Renewable Energy Microgrids, Hybrid Renewable Energy Harvesting, Energy Storage, Sustainable Business Models, Community-Based Energy Systems, Clean Energy Entrepreneurship.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Due to large availability of data in finance artificial intelligence has helped in more accurate ways to stock predication. But due to rapid and unpredictable changes in stock market has made it a challenging task. This paper gives us a intelligent stock market prediction that uses Long ShortTerm Memory(LSTM) and deep learning networks to predict the stock prices using the past financial data. The system uses real-time data from the past and online ,a machine-controlled preprocessors and user friendly Django-based web interface for comparison of actual versus predicted values. The results of this project show us accuracy and user management, showing the usefulness od the AI models like LSTMs in this project. This project stress on the capability of the AI-driven models in helping the future predication of the finance and its decisions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

accuracy , LSTMs , AI-driven models , finance , stock

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid expansion of student enrollment and recruiting partnerships has transformed campus placement into a significant logistical challenge for educational institutions. Conventional manual processes--characterized by fragmented data handling and tedious eligibility verification--are increasingly prone to administrative bottlenecks and data inconsistencies. This paper proposes a "Smart Campus Placement System" (SCPS) designed to digitize and automate the recruitment lifecycle. By leveraging a centralized database, the system integrates student academic records, technical proficiencies, and personal profiles with real-time corporate requirements. Key features include automated eligibility filtering and a streamlined communication interface between stakeholders. Preliminary findings suggest that the SCPS significantly reduces administrative overhead, ensures data integrity, and enhances the placement efficiency for both students and recruiters.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Machine learning, campus, placement

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract This study explores the presence of short story elements in Sangam literature, one of the earliest forms of Tamil literary expression. Storytelling has long been an integral part of human culture, serving as a means of communication, entertainment, and emotional expression. Sangam texts such as Akananuru, Kurunthogai, Natrinai, and Kalithogai reveal significant features of short stories, including plot, characters, dialogue, and vivid situational depiction within a concise poetic structures. The short story as a formal literary genre emerged in the modern era, particularly in Western literature and its fundamental elements can be traced back to ancient Tamil Sangam poetry. This paper analyses the selected poems to highlight the ways they encapsulate the complete narrative experiences in a brief format. Themes such as love, relationships, social values, emotional conflicts, and moral reflections are presented through engaging and thought-provoking mini-narratives. The study also illustrates the process that the Sangam poems function as a "time mirror," reflecting the lifestyle, cultural practices, and ethical values of ancient Tamil society. Through selected examples, the paper demonstrates that these poems possess the essential qualities of modern short stories brevity, intensity, and the ability to provoke deep reflection in which it establishes the continuity of the storytelling traditions from ancient to modern times.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Sangam Literature, Short Story Elements, Narrative Structure, Characterization, Emotional Themes Akananuru, Kurunthogai, Love Themes, Cultural Life, Dialogue, Storytelling Tradition

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Conducting trustworthy examinations in fully remote settings remains one of the most unresolved problems in contemporary digital education. Without a physical supervisor present, candidates gain opportunities to exploit resources, seek external assistance, or misrepresent their identity -- undermining the credibility of the entire assessment process. This paper introduces ProctoGuard, an intelligent web-based examination platform that addresses these vulnerabilities by weaving together artificial intelligence, real-time behavioral sensing, and automated learning support into a cohesive system. Rather than relying on a single monitoring signal, ProctoGuard operates across three simultaneous channels: a computer vision channel that applies facial landmark tracking to confirm candidate identity, detect unauthorized individuals within the examination frame, and observe lip movement as evidence of spoken communication [12], [14]; an acoustic channel that samples the candidate's surrounding environment through the device microphone and applies frequency analysis to distinguish human conversation from ambient noise [10], [18]; and a browser-event channel that intercepts application-switching, page-reload, and external-navigation actions the moment they occur [22]. Infractions detected across any channel are logged against the session in real time, with mild violations prompting on-screen warnings and repeated or serious violations triggering automatic finalization of the examination -- removing further opportunity for dishonest conduct without manual intervention [16]. Once an examination concludes, an automated scoring engine evaluates responses immediately and an AI-driven explanation module powered by the Google Gemini API [3] produces question-level conceptual feedback tailored to each candidate's submitted answers, converting the post-exam review into a structured learning opportunity. The platform is developed using Python and the Django framework, with MySQL providing relational data storage and Apache handling web server deployment through XAMPP [17]. Testing conducted in a controlled simulation environment confirmed an overall malpractice detection rate of 94.8%, with browser-level violations intercepted at 100% accuracy and visual and acoustic violations detected at rates ranging from 85% to 97%. All core platform operations completed within two seconds, demonstrating responsiveness suitable for live institutional use. The layered, component-based architecture of ProctoGuard ensures that the system can be extended, scaled, and maintained with minimal disruption, making it a practical and affordable option for academic institutions seeking a self-hosted solution to the challenge of secure remote assessment [15], [25].

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Remote Examination Security, Intelligent Proctoring, Behavioral Violation Detection, Facial Landmark Analysis, Acoustic Monitoring, Browser Event Tracking, Automated Assessment, AI-Generated Feedback, Django, Academic Integrity

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rise of e-commerce and digital platforms has reshaped and explored the perception of consumers to interact with FMCG products. Branding is essential for companies to invest in marketing strategies to enhance consumers' perceptions and face the competitive market in FMCG products. A descriptive research design and random sampling method are used with 102 respondents from Structured with close-ended questions. It concludes that the authorities focus on maintaining a strong brand image through the quality of the products putting effort into marketing strategies and ensuring the networks for the availability of the product to improve consumer perception.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Consumer perception, brand reputation, price, availability, strategies

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This study examines the impact of parental involvement and self-efficacy on the problem-solving ability of secondary level students. Recognized as a key 21st-century skill, problem-solving ability is essential for academic success and effective decision-making. Despite extensive research on academic achievement, limited studies have explored the combined influence of parental involvement and self-efficacy on problem-solving ability. A descriptive survey method was employed on a sample of 120 secondary students selected through purposive and random sampling from CBSE schools in Agra. Standardized tools were used to measure parental involvement, self-efficacy, and problem-solving ability. Data were analyzed using mean, standard deviation, and ANOVA techniques. The findings indicate that both parental involvement and self-efficacy significantly influence problem-solving ability, with a notable interaction effect. The study highlights the importance of supportive home environments and strong self-belief in enhancing students' cognitive skills, offering implications for educators and policymakers.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Parental Involvement, Self Efficacy, Problem Solving Ability, Academic Achievement

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The improvement in modern lifestyles has increased the need for maintaining safe and comfortable indoor environments. People are now more aware of how environmental conditions influence health, productivity, and overall well-being. Therefore, continuous monitoring of residential environments has become essential. This study presents the development of an intelligent monitoring system that utilizes Internet of Things and Cloud Computing technologies. The system collects real-time data using sensors and transmits it through IoT networks for further processing in a cloud platform. This approach enables efficient storage, analysis, and remote access to environmental data. A structured system design is proposed, including hardware components, software configuration, and data integration techniques. The system is implemented in a residential setting, where environmental parameters are monitored over time. The use of a fuzzy-based data fusion method improves measurement accuracy, maintaining error levels within 0 to 0.01. Additionally, the system evaluates indoor comfort using a standardized index, ensuring reliable and intelligent monitoring.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Internet of Things, Cloud Computing, Environmental Monitoring, Smart Home, Sensor Integration, Data Analysis.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Floods are among the most destructive natural disasters, causing severe loss of life, large-scale property damage, and disruption to agriculture, transport, and public infrastructure in many regions of the world. Many flood-prone areas, especially rural and developing regions, lack affordable and reliable systems that can continuously monitor rising water levels and intense rainfall in real time. This paper presents Innovative Flood Guard, a low-cost, embedded, standalone flood early-detection system that continuously monitors water level and rainfall using an ESP32-based microcontroller, a waterproof ultrasonic sensor, and a resistive rain sensor. Real-time sensor readings are filtered, compared against predefined safe, warning, and danger thresholds, and mapped to a three-level alert logic that drives local visual and audible alarms through LEDs, a buzzer, and an optional high-decibel siren. The system is powered by a solar-assisted battery supply and enclosed in a weather-proof housing for long-term outdoor deployment. Experimental testing under controlled conditions demonstrates that the prototype measures water level with good accuracy, distinguishes between light and heavy rainfall, and triggers alerts almost instantaneously once thresholds are crossed, making it suitable for localized, real-time flood monitoring in resource-constrained environments. The modular design also allows future integration with IoT services such as cloud dashboards and messaging applications for remote notifications and long-term data logging, while keeping the core unit fully functional in offline mode.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Flood monitoring, early warning, ESP32, ultrasonic sensor, rain sensor, embedded system, IoT-ready design, standalone alerting

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Introduction: Phyto-pectin-based hydrogels have emerged as promising biomaterials for advanced dermatological applications due to their biocompatibility, biodegradability, and bioactive potential. The present study focuses on the formulation and evaluation of a phyto-pectin hydrogel as a dual therapeutic system for skin healing and depigmentation. Methods: Pectin extracted from plant sources was combined with selected phytochemicals rich in phenolics and flavonoids to develop a stable, flexible, and moisture-retentive hydrogel. The formulated hydrogel will be assessed for favorable physicochemical properties, including optimal pH, swelling behavior, characterization and cell line assay to determine its suitability for topical application. Results: In vitro assessments is expected to demonstrated enhanced wound-healing activity through improved cell proliferation, migration, and collagen deposition, along with effective antioxidant and anti-inflammatory effects. Additionally, the hydrogel showed significant depigmenting potential by inhibiting tyrosinase activity and reducing melanin synthesis. Conclusion: The synergistic action of phyto-derived bioactives and the pectin matrix supports sustained release and localized therapeutic effects. Overall, the phyto-pectin hydrogel represents a multifunctional, eco-friendly, and safe platform with strong potential for use in skin repair and depigmentation therapies.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Plant extract, Hydrogel, Antioxidant, Characterization, Cell line assay

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

It is challenging for traditional computing methods to interpret biological data since it is inherently high-dimensional, complex, and ambiguous. Traditional approaches' accuracy and reliability are sometimes limited by things like the nonlinear dynamics of biological systems, noise in experimental data, variability in gene expression, and a lack of clinical records. Soft computing's tolerance for imprecision, ambiguity, and partial truth makes it a helpful paradigm to address these problems. Techniques such as hybrid models, fuzzy logic, evolutionary algorithms, and artificial neural networks offer flexible frameworks for analyzing a variety of biological data and spotting important patterns. The use of soft computing in systems biology, proteomics, genomics, disease prediction, and drug development is highlighted in this paper, which looks at how it can handle the ambiguity and complexity of biological data. Case studies demonstrate the application of soft computing approaches in gene expression analysis, protein structure prediction, and medical decision support systems. A contrast with older methods demonstrates the adaptability and robustness of soft computing in addressing complex biological concerns. The paper's conclusion discusses the limitations, integration with big data frameworks, and potential avenues for future research, emphasizing the growing significance of soft computing in the development of customized medicine and computational biology

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Soft Computing, Biological Data, Uncertainty, Fuzzy Logic, Computational Biology

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Agriculture plays a crucial role in the economic development of many countries, especially in rural regions where farming is the primary source of livelihood. However, the traditional agricultural marketing system involves multiple intermediaries such as wholesalers and retailers, which reduces farmers' profits and increases the final price for consumers. Farmers often face challenges such as price fluctuations, delayed payments, limited market access, and post-harvest losses. At the same time, consumers may not always receive fresh products at reasonable prices. The project titled "Farms to Home - Products Directly from the Farmers" proposes a digital platform that connects farmers and consumers directly through a mobile application. The system eliminates middlemen and provides a transparent marketplace where farmers can register, upload product details, set prices, and manage orders. Consumers can browse available farm products, place orders, make secure online payments, and receive home delivery. The system is developed using Android technology with a cloud-based database to ensure real-time data access and secure transactions. It includes modules for farmers, customers, and administrators to efficiently manage products, orders, and payments. The platform aims to improve farmers' income, provide fresh and affordable products to consumers, reduce post-harvest losses, and promote digital transformation in agriculture. Overall, this project offers a sustainable and technology-driven solution to strengthen the direct farm-to-consumer supply chain

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Farm-to-Home System, Direct Farmer-Consumer Connectivity,Digital Agricultural Marketplace

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Non-monetary benefits refer to rewards and advantages provided to employees that do not involve direct financial payment, addressing employees' psychological, emotional, and social needs. Employee performance reflects how effectively an employee carries out job duties and responsibilities to achieve organizational goals. This study examines the impact of non-monetary benefits on employee performance at Hiranmayee Rubber, Madurai. Data was collected from 100 employees using a structured questionnaire and analysed using descriptive statistics, correlation analysis, and regression analysis through SPSS. The results reveal a strong positive relationship between non-monetary benefits and employee performance (r = 0.676, p = 0.000), with non-monetary benefits explaining approximately 45.6% of the variation in employee performance. The findings confirm that promoting non-monetary benefits such as recognition, career development, work-life balance, job autonomy, supportive leadership, and a positive work environment can significantly improve employee performance and organizational effectiveness.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Non-Monetary Benefits, Employee Performance, Recognition, Career Development, Work-Life Balance, Hiranmayee Rubber.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper presents a personalized AI-driven chatbot, alongside a corresponding personalized application framework to foster the mental health of women by offering emotional support and estimating stress levels, besides offering contextualized responses with an empathetic approach. The system uses sentiment analysis, machine learning, and large language models (LLMs) to provide continuous and personalized interactions. A decision tree classifier is used to predict stress levels and VADER sentiment analysis and the PHQ-9 depression screening tool are used to provide additional support in the diagnosis. Rather than discussing certain stigmas and lack of access to care, the memory function of the chatbot maintains participation by storing user interaction history while providing subsequent agent interaction in a secure manner. Through the use of LLMs such as Mistral, scikit-learn, flaks, and python, this system is shaping up to facilitate better handling of mental well-being for women, as well as aiding in early detection of mental health issues.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Customized Health Embassy, Women Mental Health, Sentiment Analysis, Stress Prediction, PHQ-9, Large Language models, Memory Module, AI Chatbot.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract The study was carried out to understand the distribution patterns and habitat preferences of Spotted Deer (Axis axis) within the Kuvempu University campus, situated in the Bhadra Tiger reserve, Shivamogga District, Karnataka, India. Employing a multi-method approach, field surveys were conducted for a span of 6 months from January 2025 to June 2025, with methodologies like, line transect sampling, camera trap deployment, and indirect sign identification, across diverse habitat types. Out of 1247 total photographic captures from the camera trap dominated ones were the spotted deer with 26%. The density was high along the semi disturbed habitat with 14.15 compared to forest and the disturbed habitat. Group size varied from 1 to 22 individuals. Camera trap analysis documented peak activity periods during early morning from 05:00-08:00 hours and late afternoon from 04:00-06:00 hours, consistent with crepuscular behaviour patterns. Threat assessment revealed the presence of potential predators like tiger, leopard, dholes and the major effect was observed from the stray dogs. The findings reveal a pronounced association between habitat heterogeneity and Spotted Deer presence and abundance. As the spotted deer has the high ecological significance as a keystone herbivore species and primary prey base, conservation of healthy Spotted Deer populations contributes substantially to broader landscape-scale ecosystem integrity. These results underscore the critical importance of maintaining structurally complex, heterogeneous habitats and minimising disturbances for the effective conservation and management of this keystone ungulate species in shared landscapes.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Keywords: Axis axis, chital, camera trapping, line transect, habitat preference.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Illicit arms trafficking shows up as one of the biggest danger to national and international security in the today's era. Estimated cautiously between $1.7 billion and $3.5 billion within one year, the illegal trade in small arms and light weapons (SALW) propels Rebellions, violence, Criminal syndicates, and Hostilities all across the world. India, with its strategic location in the heart of South Asia and closely bordered by unstable states and Penetrable borders, faces severe problems from this Happening . This research paper tries to do a all round synopsis of India's statutory, departmental, Functional and Active responses to illegal arms trafficking along with comparing it to contemporary global legislation, including the UN Arms Trade Treaty (ATT), the UN Programme of Action on Small Arms and Light Weapons (UNPoA), the Firearms Protocol, and the UNTOC. It tries to check on statutes principally the Arms Act, 1959 (as amended in 2019), the UAPA, and the Bharatiya Nyaya Samhita, 2023, relevant from India's domestic view point and investigates on the loopholes that exists in statutory recommendation and real life circumstances. The paper suggests on how despite having a strong legislative framework, robust institutional support, brave border security, and digital surveillance system, India still need a new, comprehensive and more grounded plan to secure it's National integrity and security along with global interactions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Illicit Arms Trafficking; Small Arms and Light Weapons (SALW); National Security; Transnational Organized Crime; Border Security

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Index Terms--CO2 Emission Prediction, Machine Learning, Regression Models, XGBoost, LightGBM, CatBoost, MLP Regressor, Streamlit Dashboard, Climate Change, Data Analytics.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The increasing complexity of modern software development and deployment has led to an urgent need for intelligent solutions to predict and prevent system failures. DevOps practices, which emphasize continuous integration, continuous delivery, and automated testing, have significantly improved software reliability. However, unpredictable failures continue to impact system stability, deployment efficiency, and user experience. Artificial intelligence (AI) has emerged as a powerful tool to enhance failure prediction and prevention by leveraging machine learning models, anomaly detection techniques, and predictive analytics. This article explores how AI-powered solutions are revolutionizing DevOps by predicting failures before they occur, identifying root causes, and automating preventive measures. It examines the challenges faced in integrating AI into DevOps workflows, such as data quality, model interpretability, and system scalability. The study also discusses real-world applications, methodologies for implementing AI-driven failure prediction, and future research directions to further optimize AI's role in DevOps.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

AI in DevOps, failure prediction, failure prevention, machine learning, anomaly detection, predictive analytics, DevOps automation.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The Transformer architecture, originally developed for Natural Language Processing, has emerged as a prominent new approach in computer vision. This review concentrates on Vision Transformers (ViTs) and their application in medical imaging. We discuss the fundamental Transformer components, including self-attention, multi-head attention, and encoder-decoder structures, then describe how images are divided into patch-based sequences for transformer processing. The review highlights key elements of ViT architecture, such as patch embedding, positional encoding, encoder design, and feed-forward layers, and compares ViTs to Convolutional Neural Networks (CNNs) regarding feature extraction, local versus global context understanding, efficiency, and benchmark results. We also examine hybrid models that combine CNNs' emphasis on local features with the broad modeling capacity of self-attention. Additionally, we review notable ViT models like the Swin Transformer, DeiT, and those tailored for medical tasks like TransUNet and UNETR. Our findings indicate that, although CNNs remain effective with limited data, ViTs and hybrid models generally outperform pure CNNs in large-scale, resource-intensive medical imaging tasks. These outcomes highlight the increasing importance of Vision Transformers in medical applications such as tumor detection, organ segmentation, and disease classification.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Vision Transformers, Medical Imaging, Self-Attention, Convolutional Neural Networks, Swin Transformer, DeiT, Patch Embedding, Deep Learning.

  License

Creative Commons Attribution 4.0 and The Open Definition