VOLUME 14, ISSUE 4, APRIL 2025

Abstract: Integration of security automation tools within the Software Development Life Cycle (SDLC) is important to enhance the security posture and establish a Secure Software Development Lifecycle. We have reviewed existing research papers, articles and identified gaps in them and tried to reduce and mitigate those gaps with our proposed solution Tools like SonarQube and Dependency check can be integrated with CI/CD pipeline and help in identifying security vulnerabilities early in software development lifecycle. GitHub is source code management and version control tool, which also helps in automation of the code merge and review process. Results of this scan will be uploaded in Defect Dojo, which is an open-source tool by OWASP. Defect Dojo will serve as a central vulnerability management solution. Proposed solution in this paper will help in achieving increased detection of vulnerabilities, reduction in manual effort and a better collaboration between engineering teams and security teams. The goal of this research is to offer a solid framework for incorporating security automation into the SDLC, utilising the advantages of different tools to improve security procedures by facilitating early detection and lower risks.

Keywords: DevSecOps, Security Automation, SAST, Secure SDLC, Security Integration, SonarQube, Continuous Security Assessment

Abstract: The integration of the Internet of Things (IoT) in agriculture has transformed conventional farming by enabling real-time monitoring, data-driven decision-making, and automation. This research explores the implementation of IoT-based smart agriculture systems aimed at enhancing productivity, optimizing resource utilization, and addressing critical challenges such as water management, soil health monitoring, and crop protection. By utilizing sensors, wireless communication, and cloud computing, these systems offer farmers valuable insights into environmental conditions, facilitating precision farming techniques that enhance yield quality and efficiency. This paper delves into the architecture of IoT-enabled smart agriculture, emphasizing essential components such as soil moisture sensors, water level sensors, and automated irrigation systems. Additionally, it analyzes the benefits and challenges of adopting IoT in agriculture, including cost considerations, technical barriers, and the necessity for user-friendly solutions tailored to farmers. The study concludes by identifying emerging trends in IoT-driven agriculture and highlighting its potential to promote sustainable farming through advanced technological integration.

Keywords: Automated Irrigation Systems, Environmental Monitoring, IoT in Agriculture, Precision Agriculture, Smart Farming, Sustainable Farming.

Abstract: Agriculture is a key industry for world food security, but crop diseases are a major threat to agriculture productivity. Early and precise detection of diseases is critical to avoid loss of yield and achieve a sustainable agricultural system. Most of the farmers, particularly in India, have a hard time with disease diagnosis because they lack proper infrastructure, which results in improper management of the crops and lower yields. Machine intelligence and advanced learning provide cutting-edge techniques for detecting diseases early through computer vision approaches. In our research, we created TomatoShield, an app based on a mobile platform that has the capability of disease identification. We tested CNN models like Xception for classifying tomato plant disease using a 22,200 images dataset of leaves from Kaggle. The Xception model had achieved the accuracy of 93.64%. The app, developed with Python's Kivy library, allows farmers to take or upload images of leaves for immediate diagnosis. It also includes storing results in an SQLite database for easier recovery and analysis, delivering actionable information to farmers to increase agricultural productivity and crop health.

Keywords: Plant Disease Prediction, Xception, CNN, Disease Treatment, Image Classification, Machine Learning, Deep Learning.

Abstract: This project intends to develop a comprehensive categorization system based on machine learning to address the complex concerns of online harassment and discrimination on social media platforms. Inspired by recent research advocating for the use of machine learning in social media moderation, this project builds on existing methodologies to create a comprehensive framework capable of identifying various types of harmful content, such as cyber and non-cyber bullying, as well as discrimination based on ethnicity, gender, age, and religion. Machine learning models such as Naive Bayes, SVM, Random Forest, Decision Tree, and Sklearn classifiers are trained to detect patterns and subtle nuances indicative of online abuse and discrimination by utilizing diverse datasets representing instances of harmful behavior across multiple dimensions. The suggested categorization system's performance and flexibility are tested by comprehensive testing and assessment on real-world social media data. The technology provides timely and precise identification of hazardous information by combining different categorization tasks under a uniform framework, allowing social media platforms to handle its propagation proactively. Furthermore, the use of machine learning algorithms improves the scalability and effectiveness of content moderation activities, reducing the burden on human moderators and creating a safer and more inclusive online environment. This study contributes to a better understanding of the complex dynamics of online abuse and discrimination, enabling the creation of nuanced solutions for enhancing online safety and content control.

Keywords: child predators, cyber harassers, Twitter, machine learning.

Keywords: AI in Healthcare, Medical Chatbot, E-commerce, Patient Management, Automation.

Abstract: Using data-driven technologies, modern agriculture promotes sustainability while increasing farming efficiency. Intelligent farming solutions, like the one suggested in this paper, are now possible thanks to developments in machine learning (ML) and the internet of things (IoT). Using an Internet of Things framework, it integrates AI-based crop recommendations with real-time soil monitoring. Together with a temperature and pH sensor, a multi-electrode NPK sensor measures the three main nutrients found in soil: nitrogen (N), phosphorus (P), and potassium (K). An ESP32 microcontroller receives data through an RS485-to-USB interface, processes it, and shows it on a 20x4 LCD I2C screen. Real-time weather data from API integrations is also added. Reliance on conventional lab testing is decreased by cloud-based machine learning models that evaluate environmental variables and soil health to recommend appropriate crops. Because of the system's remote accessibility, farmers—even in remote locations with little technical know-how—to make wise choices. Additional sensors (such as light and moisture sensors) and weather-adaptive predictive models are examples of future improvements that could improve the accuracy of the system and encourage sustainable farming methods.

Abstract: Our project ultimately aims to offer a scalable and flexible solution that can be customized to meet the specific needs of various industries, from finance to healthcare. By ensuring compliance with data privacy regulations and implementing a robust security framework, our project not only protects sensitive data but also helps organizations build trust with customers and regulatory authorities. Its comprehensive approach ensures that data management practices align with both current and future data protection requirements. With digital connectivity at an all-time high, protecting personal information has become a matter of utmost concern for individuals, organizations, and governments around the world. This paper delves into the complex world of International Data Privacy Regulation (IDPR) and offers a comprehensive solution to tackle global data protection issues. An Electronic Health Record (EHR) is an electronic copy of a patient's paper chart, intended to hold a complete, up-to-date, and patient-focused set of health data. It is utilized by clinics, hospitals, and healthcare providers to coordinate care efficiently.

Keywords: Data Privacy, React.Js, Node.Js, MongoDB, Axios, Express.js, Tailwind CSS, Bcrypt

Abstract: In present scenario, people are buried up in a heavy work culture, as everyone is engaged with busy schedules, and hectic tasks which make them deviate from family life. If any issues encounter unexpectedly, it distracts them and makes them choose over the work they have to accomplish primarily. It is important to manage both professional and family life. In such circumstances, every one of us would have fantasized about a kind of house which doesn’t have any leaks in pipes, if it doesn’t have any mess in fixing a furniture and a kind of house which never face any maintenance issues and every one of us have thought that a life would be much better if no point of issue arises in getting a service at your door step and if there is no mess in bargaining a labor for home service. In such situation’s E-Commerce plays a vital role in today’s life as it has so many advantages in our life because it makes convenient in daily life of the people.

Keywords: Urban ease, home services, customizable packages, AI-based recommendation, secure payment gateways, user experience, operational efficiency

Abstract: Early detection of plant diseases is essential for safeguarding crop health and enhancing agricultural productivity. Traditional methods relying on manual observation are often slow, inaccurate, and inefficient. This ponder proposes a unused approach utilizing Convolutional Neural Systems (CNNs) to consequently distinguish plant maladies, altogether upgrading speed and precision. The show is prepared on a comprehensive collection of pictures, covering both solid and infected plant clears out, driving to a tall discovery rate. By joining exchange learning, the framework can perform successfully indeed with constrained information. Planned to function in real-time and at scale, this device is available to agriculturists, advertising a viable arrangement that diminishes the require for pesticides, bolsters way better trim administration, and empowers more feasible rural hones.

Abstract: The QR Code-Based Student Attendance System is a cutting-edge solution for automating and simplifying attendance management at schools. The conventional method of taking attendance using manual roll calls and paper registers takes too much time, is prone to errors, and is susceptible to proxy attendance. This project leverages QR code technology to offer a secure, efficient, and contactless method of capturing student attendance. Every student receives a personalized QR code that contains their identification information. In the classroom sessions, the codes can be scanned by the teachers through a webcam or smartphone, and the attendance is automatically recorded in a central database. Implemented in Python with libraries such as qrcode, OpenCV, and pyzbar, the system provides role-based access for administrators, teachers, and students and offers functionalities such as monthly attendance reports and processing data in real time. The solution ensures higher accuracy, less manual effort, and digital transformation within academic settings.

Keywords: QR Code, Attendance Management, Automation, Student Monitoring, Python, OpenCV, Pyzbar, Real-Time Processing, Role-Based Access, Data Logging, Educational Technology, Contactless System, Proxy Prevention, Centralized Database, Monthly Reports

Abstract: Malicious behaviour analysis is a critical aspect of cybersecurity aimed at identifying harmful activities such as data exfiltration, privilege escalation, and system exploitation. Traditional methods often rely on predefined signatures or shallow heuristics, which limit their ability to detect evolving or previously unseen threats. To address these limitations, this study employs a deep learning-based approach utilising Vanilla Transformers, a model architecture renowned for its powerful attention mechanisms and ability to capture complex dependencies in sequential data. Unlike recurrent architectures, Vanilla Transformers process entire sequences in parallel, enabling faster computation and more effective learning of behavioural patterns. The model demonstrated strong performance, achieving 99.89% accuracy, 100% recall, 100% precision, and an F1-score of 100%, indicating its effectiveness in identifying malicious behaviours with minimal false positives. This research highlights the potential of attention-based architectures in cybersecurity, providing a scalable and adaptive solution for real-time threat detection and behavioural analysis in complex digital environments.

Keywords: Malicious Behaviour Detection, Network Intrusion Detection System (NIDS), Vanilla Transformers, Deep Learning, Cybersecurity, Wireshark, CiscoFlow Meter, Real-Time Network Monitoring

Abstract: Access to essential medicines during emergencies remains a significant challenge due to limited pharmacy availability and accessibility constraints. Our project introduces an Automatic Drug Dispenser, a smart, self-service system designed to provide quick, secure, and 24/7 access to both prescribed and over-the-counter medicines. The dispenser integrates QR-based prescription verification, digital payment options, and real-time inventory tracking to ensure seamless medication dispensing. to enhance accessibility further, our system features online telemedicine services, allowing users to request medicines remotely, consult with licensed doctors via video calls, and receive digitally verified prescriptions. This hybrid model ensures that even patients without prior prescriptions can access the right medication through on-the-spot online consultation. targeted for deployment in hospitals, highways, universities, and remote areas, our solution minimizes dependency on traditional pharmacies, enhances emergency healthcare response, and bridges the gap between patients and healthcare providers. By integrating automated dispensing with telemedicine, our project aims to revolutionize medication access, prescription management, and remote healthcare support.

Keywords: Automatic Drug Dispenser, QR-Based Prescription Verification, Digital Payments, Real-Time Inventory Management, Telemedicine, Online Medicine Request, Smart Vending Machine, Pharmacy Automation, Prescription Management.

Abstract: In order to provide information about medicinal plants, this research presents a unique approach that combines handwriting recognition and image categorization with a chatbot. The project utilizes the Inceptionv3 algorithm for image classification to identify leaves and the medicines derived from them. For handwritten recognition, Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks are employed to identify medicine names. An Al chatbot is then used to resolve user queries about these medicines and provide additional details such as home remedies. The project addresses the problem of efficiently retrieving information about medicinal plants and their uses, which is challenging due to the vast amount of information. available and the need for accurate identification. Existing approaches often lack the integration of multiple technologies or focus on a specific aspect of the problem. This work contributes by providing a comprehensive solution that combines image classification, handwritten recognition, and chatbot technology. The significance of this work lies in its potential to improve access to information about medicinal plants, which can have a significant impact on healthcare and the environment. The project was conducted over a period of November-February at Guntur and has the potential to benefit a wide range of stakeholders, including healthcare professionals, researchers, and individuals interested in natural remedies.

Keywords: Medicinal plants, chatbot, image classification, handwritten recognition, CNN, herbal medicine, natural remedies, personalized recommendations.

Abstract: A Decentralized Storage Network (DSN) that combines blockchain technology with peer-to peer (P2P) communication to enable secure, tamper-proof file storage across distributed nodes. This system ensures data confidentiality and integrity by splitting files into encrypted chunks, redundantly storing them across the network, and recording storage commitments on an immutable blockchain ledger. This System achieves P2P distributed storage system which does not have single point of failure, and tamper-proof file storage with intuitive design and ease of access. The data chunks are duplicated and stored on multiple nodes for fault tolerance and leveraging Proof of storage consensus for Security and Validating Nodes and rewarding them. This network include a lightweight Merkle root-based transaction validation mechanism, UDP hole-punching for NAT traversal, and AES-256 encryption for data confidentiality.

Keywords: Decentralized Storage Network (DSN), Blockchain-based Storage, Peer-to-Peer (P2P) Networking, Tamper-proof File Storage.

Abstract: This paper presents design of reaching law based intelligent Sliding Mode Controller to improve the control of 2 degree of freedom serial robotic joints. Sliding Mode Control is very prominent control structure for robust control of the nonlinear systems. In this study, Radial Basis Function based intelligent sliding mode control is designed. This novel Control structure has two important components. First one is reaching laws that ensured finite-time convergence of the system trajectories to the sliding surface while reducing the chattering and second part of the control structure is an adaptive Radial Basis Function Neural Network that addresses the unknown system dynamics and external disturbances. The combined approach improve the accuracy, stability, and robust under model variations. The systems closed loop stability analysis is guaranteed by Lyapunov stability and adaptive laws are derived for online updating of neural network parameters. The proposed intelligent structure is implemented to robotic joints and simulation results demonstrate the effectiveness of the proposed control structure by achieving more accurate tracking, low control effort, and better disturbance rejection compared to conventional sliding mode control.

Keywords: Sliding Mode Control, Reaching Law, Radial Basis Function, Robust Control.

Abstract: In the era of digitization, artificial intelligence has revolutionized the way we process and analyze data. However, a significant portion of historical documents and cultural heritage remains in handwritten form, inaccessible to digital technologies. Optical Character Recognition (OCR) emerges as a crucial solution, enabling the conversion of optical text into digital format, thereby making it editable, searchable, and electronically storable. This technology is vital for organizations and individuals dealing with vast amounts of textual information. By training OCR engines on diverse languages, including Telugu, we can tap into the rich cultural heritage of India’s classical languages. Telugu OCR, in particular, facilitates the preservation of hand- written notes, ancient manuscripts, and historical documents, making them accessible to a broader audience. This digital transformation not only preserves cultural heritage but also enables the dissemination of knowledge and ideas to a wider audience, promoting cultural exchange and understanding.

Keywords: Handwritten Telugu Character Recognition, Optical Character Recognition (OCR), Neural Networks, Deep Learning, Image Processing, Pattern Recognition, Machine Learning

Abstract: As digital education continues to rise, the demand for reliable online exam monitoring has grown significantly. Conventional in-person invigilation methods are not feasible in remote settings, highlighting the need for automated proctoring solutions. This paper presents a web-based proctoring system that utilizes real-time webcam monitoring through WebCam.js and server-side functionality powered by Node.js. Instead of relying on artificial intelligence or machine learning, the system adopts a straightforward rule-based approach to identify suspicious behaviors such as switching browser tabs, inconsistent facial presence, and unusual eye movements. A tiered warning system is implemented, where repeated violations lead to automatic termination of the examination session. Built on the MERN stack, the platform emphasizes scalability, user accessibility, and exam integrity. System evaluations indicate high reliability in detecting anomalies, making it an effective alternative to manual supervision.

Keywords: Remote Proctoring, Online Exams, WebCam.js, Node.js, Eye Tracking, Academic Honesty, Cheating Detection.

Abstract: Artificial Intelligence (AI) is revolutionizing urban planning by enabling data-driven decision-making and enhancing sustainability. This paper explores the transformative potential of AI in optimizing urban design, improving resource alloca- tion.The integration of AI technologies in urban planning not only streamlines processes but also addresses complex challenges such as traffic congestion, environmental sustainability, and social equity. By harnessing vast datasets, AI can identify patterns and predict future urban trends, allowing planners to make informed decisions that enhance the quality of life for residents.The role of AI will be crucial in shaping urban landscapes that are not only efficient but also equitable and responsive to the needs of diverse populations. In conclusion, this research aims to provide a comprehensive framework for understanding the impact of AI on urban planning, highlighting best practices and innovative approaches that can lead to sustainable urban development in the 21st century.

Abstract: Moisture content is a key factor affecting the quality, shelf life, and market value of grains. Accurate detection is essential for proper storage, processing, and overall post-harvest management. This study reviews both traditional and modern techniques for measuring moisture content. While methods like oven drying are reliable, they are time-consuming and unsuitable for real-time use. In contrast, advanced techniques such as capacitance, microwave, and near-infrared (NIR) spectroscopy provide faster, more accurate, and often non-destructive alternatives. The study discusses the working principles, advantages, and limitations of each method, focusing on accuracy, efficiency, and cost. Experimental results highlight the effectiveness of modern, non-invasive techniques in delivering rapid and precise moisture readings without damaging the grain. These technologies enable better decision-making in storage and processing, helping reduce post-harvest losses, maintain grain quality, and support food security.

Keywords: USB to TTL UART Serial Converter, ADS1115, MS51FB9AE, etc.

Abstract: This study uses the IMDb movie review dataset to compare how well different deep learning architectures perform in sentiment classification. The experiments evaluate three sequential models, each enhanced with distinct configurations of activation functions and layer compositions. First Model utilizes ReLU and ELU activations within a bidirectional LSTM architecture, Second Model incorporates Tanh and SELU functions, while Third Model adopts a combination of Leaky ReLU and Tanh within a similar structural framework. To determine the effect of network architecture and activation function selection on classification effectiveness, each model is evaluated using accuracy, precision, recall, F1 score, and loss measures. Results indicate that the incorporation of advanced activations such as SELU and Leaky ReLU can lead to performance gains in certain metrics, with Third Model demonstrating improved generalization and lower loss compared to its predecessors. These results highlight how important activation functions are for improving deep learning models for tasks involving natural language processing.

Keywords: Sentiment Analysis, IMDb Dataset, Deep Learning, LSTM, ReLU, ELU, Tanh, SELU, Leaky ReLU, Natural Language Processing.

Abstract: This paper presents an innovative GPS-based toll collection system developed for the Mumbai-Pune Expressway, aimed at eliminating physical toll booths through the integration of geofencing and real-time GPS tracking. Leveraging OpenStreetMap (OSM) data processed via QGIS, the system defines virtual toll zones and calculates charges dynamically using the Haversine formula. Machine Learning models are incorporated to optimize toll pricing and enable automated remote fee processing. A web-based dashboard facilitates real-time monitoring and digital payments, enhancing user convenience and operational transparency. Experimental results demonstrate a 92% prediction accuracy, 30% reduction in computational overhead, and 40% cost savings compared to traditional tolling infrastructure. The proposed system showcases a scalable, efficient, and storage-free alternative for modernizing toll collection practices.

Keywords: GPS Tolling, Geofencing, Dynamic Pricing, Machine Learning, Smart Transportation, Cloud-based Toll Collection, Real-time GPS Monitoring.

Abstract: Quantity Surveying (QS) plays a critical role in overseeing and managing construction project costs. However, conventional QS practices—often dependent on paper-based documentation and Excel spreadsheets—can be inefficient and prone to estimation inaccuracies. This study introduces a web-based QS automation framework developed using ASP.NET, designed to improve the efficiency and accuracy of cost estimation for concrete structures. The system provides a user-friendly interface that aligns with the logical sequence of construction activities, thereby improving usability and workflow. Expert validation using real project data demonstrated the framework’s effectiveness, showing a notable improvement in accuracy and speed. The automated approach achieved a 99% accuracy rate in cost estimation, significantly reducing the computation time from 114 days (manual) and 19 days (Excel) to just 3 days. These improvements highlight substantial savings in both labor and time, offering a reliable and scalable solution to modern QS challenges.

Keywords: Quantity Surveying; Automation Framework; Concrete Construction; Web-Based System; ASP.NET; Cost Estimation; Project Management.

Abstract: Smart Go Shala is an innovative approach to modernize traditional cow-based learn ing centers by using digital tools and technology. It aims to combine ancient Indian knowledge about cows, agriculture, and natural living with smart solutions like mobile apps, digital classrooms, and interactive learning materials. Through this system, stu dents can learn about the importance of cows in Indian culture, organic farming practices, and sustainable living in a more engaging and effective way. The Smart Go Shala helps preserve our heritage while making education more accessible, interactive, and suitable for the modern world.

Keywords: OTP (One Time Password), QR Code Scanner, Web Application, smart phone applications, Mail Confirmation, Entry System, Dr. Pannel, Admin Pannel, Cow Tretment, Milk Production, Cow Health, Financial Benefits etc.

Abstract: This research addresses key challenges in digital healthcare specifically data security, privacy, and integrity of medical records—by proposing a blockchain-based system powered by the SHA-256 algorithm and machine learning. Traditional centralized healthcare systems face threats like unauthorized data access, inefficiency, and manipulation. The proposed system provides a decentralized, tamper-proof environment for storing and accessing medical records while offering predictive analytics for disease detection. Using RFID-based authentication, smart contracts, and a machine learning prediction engine, the system ensures secure and intelligent healthcare. Implementation and experimentation demonstrate improved data integrity, security, and diagnostic accuracy.

Keywords: Blockchain, Healthcare System, SHA-256, Machine Learning, Data Security, Predictive Analytics, Smart Contracts, RFID Authentication, Medical Data Privacy, Decentralized Storage, Disease Prediction, Healthcare Efficiency.

Abstract: The early and accurate detection of plant diseases is crucial for maintaining agricultural productivity and food security. This paper presents an advanced Convolutional Neural Network (CNN) architecture for classifying ten distinct tomato leaf diseases with high precision. Utilizing a dataset of 16,021 annotated tomato leaf images from the PlantVillage repository, we developed a six-layer deep CNN model that achieves superior classification performance compared to existing approaches. Our methodology incorporates extensive data augmentation, care- ful hyperparameter tuning, and a systematic evaluation across multiple training epochs (10, 20, and 50). The proposed model demonstrates progressive improvement in classification accuracy, reaching 97% at 50 epochs, with particular strengths in distin- guishing visually similar diseases like early blight and late blight. We further implement a practical web-based interface using Streamlit to facilitate real-world deployment. Comprehensive ex- periments validate our architecture’s effectiveness, with detailed analysis of feature importance and model interpretability. This work contributes to the growing field of precision agriculture by providing farmers with an accessible, automated tool for plant disease diagnosis, potentially reducing crop losses by enabling timely intervention. Index Terms: Convolutional Neural Networks, Deep Learn- ing, Tomato Leaf Disease Classification, Precision Agriculture, Computer Vision, Plant Pathology, Automated Disease Detection

Abstract: Accurate sales forecasting is essential for retail businesses to optimize inventory, enhance customer satisfaction, and drive strategic decisions. This paper introduces a robust sales prediction system that integrates Long Short-Term Memory (LSTM) networks for time-series forecasting and XGBoost for predictive analytics to deliver reliable and precise sales predictions. Designed for modern retail environments, the system seamlessly integrates with Point-of Sale (POS) systems to enable real-time data ingestion and dynamic prediction capabilities. Users can also upload custom datasets and explore interactive modules for analyzing current sales trends and forecasting future demand. A React-powered dashboard offers intuitive data visualization, while a Flask-based backend ensures scalability and efficient processing. By combining cutting-edge machine learning models with real-time data handling and user-centric features, this solution empowers retailers to respond to market changes and gain a competitive advantage proactively.

Keywords: Retail, sales prediction, LSTM networks, XGBoost, real-time fore casting, POS integration, machine learning, interactive analytics, time-series prediction, data visualization.

Abstract: It underlines the rather important role that cervical cancer plays in being a major cause of deaths from cancer among women around the world, not necessarily because of other reasons it progresses slowly and often unpredictably. Early detection through screening forms the basic steps involved in the prevention of cervical cancer and this involves identification and monitoring of precancerous zones within the cervix, which again can be categorized into three different types namely, type 1, type 2, and type 3. Proper identification and analysis of every one of these stages can effectively check their progression into invasive cancer. Such appeals for accurate classification of cervical pre- cancerous images into these categories through highly advanced automated systems. The intelligent system through artificial intelligence as well as machine learning is designed to improve efficiency and precision in the cervical cancer screening so that timely intervention is facilitated. Systems focused on the more individualized and targeted approach tend to prevent the precancerous cells from being transformed into cancerous cells. Automated tools provide a reliable alternative in resource constraint settings whereby the screening process done through manual tools is not in place, and such a screening becomes possible with fewer rates of error in diagnosing it, which also happens to be accessible since a deep model like ResNet-50 generates notable performance in the colposcopy image classification that improves the cervical cancer screening as well as the preventive measures in place. Such discoveries promise a new direction in the treatment of cervical cancer and significantly fewer deaths from that disease, therefore ensuring improved overall results for women's health worldwide.

Keywords: Cervical cancer screening, Colposcopy images, Deep learning, Diagnostic accuracy, Early detection, Machine learning, Pre-cancerous zones, ResNet-50, Screening.

Keywords: Image, Caption, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) , Long Short Term Memory(LSTM), Neural Networks.

Abstract: Crowdsourcing has become a powerful model for solving complex tasks by leveraging human intelligence and emerging technologies. This paper explores its applications in game design, blockchain, hackathons, and education, analyzing key mechanisms and challenges like quality control, security, and worker motivation. It highlights innovations such as blockchain verification, AI automation, and the Social Internet of Things (SIoT) to enhance efficiency. By synthesizing insights from multiple studies, this paper provides a comprehensive view of trends, challenges, and future directions in crowdsourcing.

Keywords: Crowdsourcing, Blockchain, Game Design, Hackathons, Social IoT, AI, Data Quality.

Abstract: This paper introduces an AI-based story generation system using Recurrent Neural Networks (RNNs) to create structured stories, such as dialogues, emotions, and cinematic scene descriptions. The proposed model employs a deep learning scriptwriting automation process with the integration of AI- enabled visuals, voice acting, and cinematic editing to create a fully automated storytelling process. The system is intended to boost creativity through the input of story prompts, which the AI transforms into well-structured stories. Using large-scale structured movie script datasets, the AI acquires patterns of storytelling, such as character interactions, scene changes, and narrative pacing. The use of Stable Diffusion for visual generation and ElevenLabs for voice synthesis further enriches the story- telling experience, creating a multimedia-rich output. The use of MoviePy also provides for audio and video integration without any gaps, creating professional-quality cinematic presentations. The ultimate aim of this system is to extend the limits of AI- assisted creativity, offering a tool for writers, filmmakers, and content creators to venture into new horizons in automated storytelling. Index Terms: Artificial Intelligence, Recurrent Neural Net- works, Deep Learning, Story Generation, Automated Scriptwrit- ing, Natural Language Processing, Cinematic Scene Generation, AI-driven Storytelling, Voice Synthesis, Visual Generation, Text- to-Speech, Neural Storyteller, Machine Learning in Creativity, Reinforcement Learning in Storytelling.

Abstract: Statistical modelling has been fundamental to Natural Language Processing (NLP), providing scalable, data-driven solutions beyond traditional rule-based methods. This paper surveys key statistical models, including n-gram models, Hidden Markov Models, and Conditional Random Fields, as well as advanced methods like Bayesian models and Latent Dirichlet Allocation, which reveal hidden structures in text. We explore their applications across tasks such as part-of-speech tagging, named entity recognition, machine translation, and text classification. The paper also reviews evaluation metrics like perplexity, BLEU, and F1-score, and discusses challenges such as data sparsity and limitations in capturing long-range dependencies. A comparison with neural-based approaches highlights scenarios where statistical models remain preferable, particularly for interpretability and low-resource settings. We conclude by recommending hybrid statistical-neural models to achieve effective, interpretable, and efficient NLP solutions.

Keywords: Natural Language Processing, Statistical Modelling, N-gram, Hidden Markov Model (HMM), Conditional Random Field (CRF), Latent Dirichlet Allocation (LDA), Sequence Labelling, Machine Translation, Language Modelling, Probabilistic Methods.

Abstract: The rise of e-commerce has made online auctions increasingly popular, but existing systems often lack robust security measures. Vulnerabilities like data breaches, unauthorized bidding, and manipulation are prevalent in traditional platforms. This paper introduces a cryptographically enhanced online auction system that ensures bid confidentiality, user authentication, and overall fairness. By integrating AES encryption, SHA-256 hashing, and role-based access control, the system guarantees the integrity of auction operations. A detailed security analysis compares our model with existing solutions, showcasing its effectiveness in addressing the security concerns of digital auction environments.

Keywords: Secure Auction, Cryptography, Data Integrity, Bid Confidentiality, Online Bidding System.

Abstract: Adherence to prescribed medication schedules and accurate identification of medicines are crucial for effective healthcare management. This paper introduces an Android-based application, Medicine Consultation Reminder and Identification, aimed at simplifying medication management for users. The application provides two primary features: a medicine identification system using Tesseract Optical Character Recognition (OCR) technology to scan and recognize medicine labels and a personalized reminder system to alert users about their medication schedules. Developed after multiple consultations with healthcare professional, the application addresses challenges such as missed doses and incorrect medicine usage. The Tesseract OCR-based medicine identification feature ensures users can verify medicine details quickly, reducing potential errors. Meanwhile, the reminder system is customizable, allowing users to schedule alerts based on their prescriptions. Additionally, the app is designed to be fully voice-command compatible, enabling hands-free operation, making it more accessible for users with mobility impairments or those who prefer voice interaction. It is also customized to assist visually impaired individuals by reading aloud medication details and reminders, ensuring they can manage their health independently. The app’s user-friendly interface and voice control features make it accessible for individuals across different age groups. Its development focuses on integrating advanced scanning technology while ensuring data privacy and usability. By bridging the gap between technology and healthcare, this project strives to improve medication adherence and support patients in managing their health. The proposed solution has the potential to enhance healthcare outcomes and reduce risks associated with medication non-compliance or errors, making it a valuable contribution to the digital health domain.

Keywords: CNN, OCR, Text-to-Speech, Cloud-based storage, Image processing, Notifications

Abstract: Facial recognition technology has been one of the most powerful tools in automating processes that included attendance management, since their emergence in recent years. This paper would present an alternative for traditional attendance management methods such as that of manual attendance sheets and card-based systems by replacing them using a web-based Attendance System assisted through Facial Recognition. The system would then be assisted with a collection of computer algorithms and image processing techniques to identify and authenticate personnel based on face features. The face capture and matching system automatically marks attendance without any human intervention with the inclusion of the camera at the entry points, capturing the faces of people and matching them with the pre-registered images in the database. The new system is efficient and convenient as well; it is secure, time-saving, and will completely rule out the scope for proxy attendance. The system can also be designed for various environments like educational institutions, corporate offices, and conferences. The performance evaluation also shows great accuracy in recognition even under changing lighting conditions and with a number of people within the frame. This results in an extremely robust, user-friendly, and scalable solution that streamlines attendance management without introducing human errors.

Abstract: Role-Playing Games (RPGs) and Massively Multi- world Open Role-Playing Games (MMORPGs) have evolved into a significant genre within the digital gaming industry, shaping interactive storytelling, social engagement, and virtual economies. This paper provides a comprehensive overview of the mechanics, design principles, and player dynamics that define RPGs and MMORPGs. Emphasis is placed on their historical evolution, key features such as character progression, quests, and immersive world-building, as well as the technological advancements that support large-scale multiplayer environments. Furthermore, the paper explores the social and psychological impacts of MMORPGs, including player cooperation, competition, and the development of in-game communities. By analyzing the intersection of narrative design, user experience, and game architecture, this study aims to highlight the cultural and technological significance of RPGs and MMORPGs in modern digital entertainment.

Abstract: The proliferation of deepfake audio poses significant challenges, including the erosion of trust in digital communications and heightened risks of fraud and misinformation. This paper presents EchoVerify, a robust detection framework integrating Mel-Frequency Cepstral Coefficients (MFCC) and Speech Emotion Recognition (SER). Using Convolutional Neural Networks (CNNs), EchoVerify extracts audio features to identify synthetic manipulations with high accuracy. Our model outperforms existing approaches in noisy conditions, making it a critical tool for applications requiring audio authentication, such as cybersecurity and digital forensics.

Keywords: Deepfake audio, EchoVerify, MFCC, Random Forest, SVM, emotion detection, audio authentication, synthetic speech, digital security, misinformation prevention.

Abstract: The Tactile Robot Interpreter (TRI) is an innovative assistive device developed to bridge communication barriers for individuals with multi-sensory impairments. By combining computer vision with robotics, TRI captures text from images or documents and translates it into American Sign Language (ASL) gestures through a robotic hand, facilitating inclusive and accessible communication. However, to ensure broader adoption and real-world applicability, there is a pressing need for a cost-effective yet fully automated solution that maintains the system's functional integrity while remaining affordable for diverse communities. The robotic hand is controlled by an Arduino Uno and actuated by servo motors, which accurately replicate ASL signs based on the corresponding text input acquired through visual methods.  A distinctive feature of TRI is its ability to convey ASL through touch, enabling blind and deaf users to perceive sign language by feeling the robotic hand’s movements. This innovation opens new avenues for communication, education, and digital accessibility, significantly enhancing the independence and social participation of the visually and hearing impaired.. By transforming text into a tangible, interactive ASL experience, the TRI project paves the way for a more inclusive and connected world.

Keywords: ASL Signs, Tactile Robot, Arduino, Text Images

Abstract: Brain age estimation is an emerging field in medical imaging, particularly useful for detecting neurological diseases and age-related cognitive decline. This project aims to develop a robust model for predicting brain age using T1-weighted MRI scans. By analyzing the structural patterns within these scans, the model will estimate the biological age of a patient’s brain. The deviation between the predicted brain age and the chrono- logical age may indicate the presence of neurological diseases such as Alzheimer’s, Parkinson’s, or other neurodegenerative conditions. The project will leverage deep learning algorithms to process MRI data and predict brain age accurately. Various preprocessing steps will be applied to ensure high-quality input for the model, and advanced neural network architectures will be utilized for prediction. The ultimate goal is to provide a tool that aids in early diagnosis of neurological conditions by identifying patients whose brains show signs of accelerated aging. This system, if effective, can enhance early detection and interven- tion strategies, improving patient outcomes and contributing to personalized healthcare solutions.

Abstract: This research investigates the application of Convolutional Neural Networks (CNNs) in identifying and managing brain tumors using MRI scans, emphasizing the importance of early and precise detection for successful treatment and enhanced patient prognosis. Conventional imaging techniques often fall short in providing reliable tumor detection, highlighting the necessity for more advanced solutions. By utilizing CNNs, this study focuses on building an accurate system capable of detecting and categorizing tumors based on distinct features, thereby improving diagnostic reliability while enabling customized treatment plans for individual patients. The methodology incorporates key factors like tumor type, size, and position to refine treatment approaches, along with ongoing monitoring to adapt therapies using real-time updates, ensuring optimal care. The project aims to make meaningful advancements in neuro-oncology by enhancing early diagnosis, facilitating tailored treatment plans, and elevating patient outcomes through cutting-edge techniques and sophisticated neural network applications.

Keywords: Brain tumor detection, MRI scans, Machine learning, Image preprocessing Convolutional Neural Networks, Logistic Regression, Personalized treatment plans, Neuro-oncology.

Abstract: This paper focuses on unraveling the inner work- ings of the transformer architecture, a cornerstone of modern enabling parallel processing and long-range dependency cap- ture. From this seminal work, w√e adopt the core attention large language models (LLMs). While transformers have driven mechanism formula (Q × K^T )/  d_k and the multi-head at- breakthroughs in natural language processing through self- attention mechanisms, their internal operations remain complex and opaque. Using GPT-2 as an illustrative case study, we develop an interactive visualization framework to map information flow, display attention patterns, and illustrate token embeddings and layer interactions. These visualizations aim to deepen compre- hension of transformer mechanics, enhance model transparency, and guide future advancements in AI design.

Keywords: Transformer Architecture (TA): Neural network ar- chitecture based on self-attention mechanisms; Large Language Models (LLMs): Advanced AI models trained on vast text datasets; Natural Language Processing (NLP): AI technology for understanding and processing human language; Self-Attention Mechanism (SAM): Method allowing models to weigh importance of different input elements.

Keywords: User Authentication, Real Time Data, Fitness Tracking, MERN Stack

Abstract: This paper introduces a Machine Learning-based detection system for Denial of Service attacks on WSNs providing robust cybersecurity to these vulnerable systems. The class imbalance problem is quite significant in the WSN-DS dataset, so SMOTE will be used to create synthetic samples to balance the distribution of instances for attack and normal data. Then, feature selection is used which guides the search for relevant attributes to effectively detect attacks. Further, three different machine learning models were trained and evaluated: Logistic Regression, Decision Tree, and REPTree, measuring them in terms of accuracy, precision, recall, and F1-score. This study illustrates that this approach works towards correctly identifying the diverse categories of DoS attacks very efficiently and creates grounds for more effective security strategies for WSNs.

Keywords: Wireless Sensor Networks, Denial of Service, Distributed Denial of Service, Machine Learning.

Abstract: This project presents a comprehensive digital plat- form designed to enhance healthcare management for both patients and doctors. The system provides secure sign-in and registration options, followed by dedicated dashboards tailored to the specific needs of each user type. Patients can access their medical records, view appointment and medication reminders, and utilize scanning tools for summarizing medical reports and prescriptions. Doctors, on the other hand, can manage patient requests, access detailed medical histories, and efficiently oversee ongoing treatments. By centralizing medical data and streamlining communication between patients and healthcare providers, the platform aims to improve the quality of care and facilitate better health outcomes. This project addresses common challenges in healthcare management, such as record-keeping and appointment scheduling, ultimately supporting a more effi- cient, accessible, and patient-centric healthcare experience.

Abstract: This project presents LuxeVogue, an AI-powered web application that provides personalized fashion recommendations based on users’ physical features. Utilizing computer vision techniques, the system analyzes uploaded images to detect skin tone and estimate body shape through keypoint detection. The skin tone is mapped to seasonal color palettes, while body measurements are used to classify body shape into categories such as hourglass, pear, rectangle, or inverted triangle. Based on these attributes and the current season, the application suggests tailored clothing and accessory styles, enhancing the user’s fashion experience. Built using Streamlit, OpenCV, and Detectron2, the system also integrates a chatbot via IBM Watson for interactive user support. This project demonstrates the potential of combining AI and fashion for intelligent style guidance. IndexTerms: LuxeVogue, Fashion recommendation system, Skin tone detection, Body shape detection, Computer vision, OpenCV, Detectron2, Augmented Reality (AR)

Abstract: Estimating Software Anomalies is a critical aspect of ensuring the reliability and quality of software projects. It involves identifying and predicting bugs or faults in the source code. By detecting faults early on, developers can address them promptly, leading to improved software development processes and reduced debugging time and costs. This project focuses on developing an intelligent system that leverages advanced machine learning techniques to predict software faults in the source code. The proposed software fault prediction (SFP) model utilizes a combination of LSTM networks, bidirectional LSTM networks, support vector machines (SVM), and neural networks. These techniques enable the system to learn complex patterns in the data and make accurate predictions. The effectiveness of the proposed system is evaluated using real-world software projects, where it outperforms existing software fault prediction models. By implementing this intelligent system, developers can quickly and accurately detect faults, leading to enhanced reliability and quality of software projects. The system's ability to identify bugs during the development stage of a new project version and detect faults in a completed version offers significant benefits. It empowers developers to improve the software development process, reduce debugging time, and ultimately deliver higher-quality software projects.

Keywords: LSTM, SFP, SVM, software anomalies, machine learning algorithm.

Abstract: The growing number of cyberattacks through malicious URLs has made automated threat detection a crucial component of cybersecurity. This paper presents a machine learning-based approach to detect and classify URLs as malicious or benign using URL-based features. We developed a web-based detection system using the Flask framework that enables users to input URLs and receive real-time threat predictions. The model is trained on a labelled dataset and utilizes features such as URL length, presence of symbols, digits, and suspicious substrings. Among several algorithms evaluated, the Random Forest classifier delivered the highest accuracy. The system architecture supports efficient integration of the model with the Flask application, ensuring minimal response time and a user-friendly interface. Experimental results demonstrate that our approach achieves a high level of accuracy, precision, and recall. This work offers a practical, lightweight solution for integrating machine learning-based URL detection into web services, browsers, or corporate gateways, thereby enhancing user safety against phishing and malware attacks.

Keywords: Malicious URL, Machine Learning, Flask, Web Application, Cybersecurity, URL Classification, Phishing Detection, Supervised Learning, Random Forest, Web Security Automation, Threat Intelligence, Feature Engineering.

Abstract: This Document centers on the development of an advanced chatbot system that seamlessly integrates with PDF documents, significantly enhancing users’ ability to extract information using natural language queries. It addresses the growing need for efficient information retrieval from textual content, particularly in academic and professional contexts. The system provides a user-friendly platform designed to quickly and accurately extract relevant information from PDFs. To achieve this, it incorporates several modern technologies. Streamlit is used to build an intuitive and interactive user interface. For PDF parsing and text extraction, the system employs PyPDF2. LangChain is responsible for text processing and generating semantic embeddings, which improve the efficiency and relevance of indexed data. Google’s Generative AI powers the chatbot, enabling it to understand complex user queries and generate accurate, context-aware responses. Additionally, FAISS is integrated to support similarity-based search, ensuring fast and precise information retrieval from the vectorized content. The system workflow begins with users uploading PDF files, which are then parsed, processed, and indexed. The chatbot interacts with users by understanding their queries and providing targeted responses based on the indexed content. The primary aim of this project is to offer a highly interactive and user-centric experience, simplifying how users engage with and extract insights from PDF documents. Future enhancements may include support for more complex queries, broader document format compatibility, and advanced features to improve user engagement. Ultimately, this project contributes to the advancement of natural language processing and intelligent information retrieval, offering value to a wide range of domains requiring effective document analysis.

Keywords: LangChain, Vector Database, Multi-PDF Chat, AI, FAISS, OCR, Streamlit, GPT.

Abstract: During their eight semesters of study, engineering students at APJ Abdul Kalam Technological University (KTU) must accrue 100 activity points. These points are awarded based on participation in workshops, internships, startups, extracurricular activities, and other engagements, with each activity credited according to KTU's prescribed guidelines. Currently, certificates serve as proof of participation and completion, which are manually uploaded by officials after submission to the respective department. To streamline this process, we propose an OCR-based system that leverages text recognition and image processing technology. This system will automate certificate verification and data entry, reducing manual effort and ensuring seamless integration with the university portal.

Keywords: Text Recognition, OCR, NLP

Abstract: A thorough characterization of the evolution process of solid tumors and the composition of circulating tumor DNA would greatly facilitate the development of therapeutic approaches. This work depicts the computational modeling approach towards unearthing the intricate relationship between spatial organization in solid tumors and the circulation tumor DNA. The study assesses how spatial organization makes a difference to the tumor cells' release into the blood flow, the incurred effect on mutational landscapes, and high resolution about the simulation of heterogeneity in circulating tumor DNA. The spatial confinement was found dramatically to affect tumor evolution and ctDNA molecular composition-very important for the design of precision medicine strategies, such as non-invasive biomarker detection and personalized therapeutic interventions. Results suggest that spatial tumor organization affects the timing of ctDNA release, and therefore could influence the sensitivity of liquid biopsy to early tumor tracking and treatment response. Integrate computational models with experimental and clinical data in order to validate predictions, refine understanding of ctDNA dynamics, and push the advance in oncology diagnostics, monitoring, and treatment toward better outcomes in personalized medicine.

Keywords: Evolution of solid tumor, Computational modeling, Comprehensive simulations, Circulating tumour DNA (ctDNA), Non-invasive diagnostics.

Abstract: This project presents an AI-enhanced version of the classic 2048 puzzle game aimed at improving user performance and engagement. By integrating artificial intelligence, the system analyzes the game board in real time and suggests optimal moves to the player. The AI is trained using heuristic strategies and gameplay data to make intelligent decisions, helping users achieve higher scores and reach advanced tiles like 2048 and beyond. Features such as AI Move, Undo, and performance tracking offer a strategic and educational gameplay experience. This project demonstrates the effective use of AI in decision-making and real time interactive applications

Keywords: 2048 Puzzle Game– A popular single-player sliding tile puzzle where players combine tiles to reach the 2048 tile. Artificial Intelligence (AI)– Technology used to simulate intelligent decision-making for optimal move suggestions. Real-Time Analysis– Continuous evaluation of the game board to provide live move recommendations. Heuristic Strategies– Rule-based techniques used to guide AI decision-making efficiently

Keywords: NoSQL Database, SQL-like Commands, Local Storage, Python Database, CRUD Operations.

Keywords: Dyslexia, Assistive Technology, Adaptive Learning, Progressive Web App.

Abstract: The MailOps-CLI brings effective email management tools through a user-friendly command line system for all users to use. A Node.js application with node-imap technology permits users to control email automation and automatic search functions with ease. Users handle email better because they can enter date parameters and include files into emails through basic instructions. The MailOps-CLI system lets users dictate their emails to the interface rather than typing them. This voice-assisted email summary tool processes important message content to transmit spoken information for people who have vision problems at no human effort. Users can check email activity instantly because a reliable logging system records all system activities. By using MailOps-CLI users can easily control emails through a quick system that saves them from dealing with challenging email organization structures.

Keywords: email handling, command line tools, AI summary creation, voice-to-text conversion, security measures, ease of access, automated email actions, phishing protection, complete data encryption and email status alerts.

Abstract: Dr. B.R. Ambedkar, renowned as the chief architect of the Indian Constitution, was also a pioneering advocate for women's rights and gender equality. This paper critically examines Ambedkar's multifaceted contributions to women's empowerment in India. It explores his interventions in legislative reforms, constitutional guarantees, and social movements. Drawing from his writings, speeches, and political activism, this study argues that Ambedkar’s feminist praxis was inseparable from his broader struggle against caste and class oppression. His legacy offers vital insights for contemporary discourses on gender equality and social justice in India.

Keywords: Dr. B.R. Ambedkar, Women Empowerment, Gender Justice, Social Reform, Caste

Abstract: Land registration is a critical aspect of agriculture and sustainable development, yet traditional systems often suffer from fraud, inefficiency, and lack of transparency. To address these challenges, this study presents GreenLand: A Secure Land Registration Scheme for Blockchain and AI-Enabled Agriculture Industry 5.0 (AgriSafe). This system integrates blockchain technology and artificial intelligence (AI) to create a secure, tamper- proof, and fraud-resistant land registry. Blockchain ensures immutability and transparency of land records, while AI models—such as Extreme Gradient Boosting (XGB) and Light Gradient Boosting Machine (LGBM)— enhance fraud detection by analyzing transactional patterns and ownership claims. Additionally, smart contracts automate verification processes, ensuring regulatory compliance and reducing human intervention. The system employs the InterPlanetary File System (IPFS) for decentralized and secure storage, further enhancing data integrity. By leveraging AI-driven fraud detection and blockchain’s decentralized security, AgriSafe significantly improves land transaction efficiency, trust, and scalability in the agricultural sector .

Keywords: Land Registration, Blockchain Security, AI- Powered Fraud Detection, Smart Contracts, Decentralized Identity Management, Secure Land Transactions, IPFS, Agricultural Technology.

Abstract: In today’s fast-paced and digitally driven world, users expect services that are quick, secure, and easily accessible. However, the printing industry has been slow to adapt, with many traditional print shops still depending on manual operations, physical document transfers, and outdated payment methods. These practices not only waste time but also pose serious risks to user data and system security. Recognizing these limitations, the Let’s Print – Smart Printing System was developed as a modern, end-to-end digital solution that transforms the way printing services operate. This paper presents the implementation of Let’s Print, a smart and secure digital printing system designed to improve the way printing services are used. Traditional print shops often face issues like time delays, security risks from USB drives, and lack of user convenience. To solve these problems, this project introduces an online platform that allows users to upload documents, pay digitally, and get real-time updates when their prints are ready. The system is built using a three-tier architecture: ReactJS for the user interface, Spring Boot for backend processing, and MySQL for safe data storage. Documents are uploaded in supported formats like PDF and DOCX, and are encrypted during the transfer process to maintain privacy. The Razorpay payment gateway is used to enable cashless, secure transactions through UPI, cards, and wallets. Firebase Cloud Messaging and email alerts help users stay updated on the progress of their print orders. The system was thoroughly tested in phases to ensure smooth functionality and user satisfaction. Overall, Let’s Print offers a faster, safer, and more convenient way to manage printing tasks, and it holds strong potential for use in educational, corporate, and public service environments.

Keywords: Smart Printing, Online Document Upload, Secure Payment, Real-Time Notification, Document Privacy, Digital Transformation, Digital Printing Platform, Web-Based Printing Application

Abstract: Automatic text classification is an important task of natural language processing (NLP) along with sentiment analysis, data organization, and spam filtering applications. Traditional methods based on bag-of-words (BOW) or TF-IDF methods often struggle to capture the relationship between words. This limitation can lead to misclassification, especially for short or ambiguous texts. The synergy of word embedding techniques for text enhancement. Word2Vec converts a word into number vectors that capture semantic similarities and relationships. By feeding these vectors to XGBoost, the model can use the rich semantic information to make further category predictions. Word2Vec captures the relationship between words, allowing the model to understand context and distinguish between words. Consider words like “king” and “queen” that have similar numbers, but “king” and “bank” have different numbers. This improves classification compared to traditional methods. The combination of Word2Vec and XGBoost can handle noisy or incomplete files better than traditional methods. Word2Vec’s dense representation helps reduce the impact of misspellings or inconsistent content, increasing the power of real-world applications. Additionally, XGBoost’s ability to handle missing values and focus on the most important features improves model interpretation. The framework can be extended to multiple registration functions, making it adaptable to a wide range of text challenges. Finally, XGBoost’s scalability ensures that the method can be effectively applied to large datasets without sacrificing performance Index Terms: XGBoost, Word2Vec, text categorization, semantic relationships, NLP, gradient boosting, word embeddings, accuracy, machine learning, document classification.

Abstract: In today’s digital age, face recognition systems have become increasingly important across various sectors. As one of the most commonly used biometric technologies, face recognition serves multiple purposes including security, authentication, and identification. Although its accuracy may not match that of iris or fingerprint recognition, it remains widely adopted due to its contactless and non-intrusive nature. At the end of each session, attendance records will be automatically sent to the respective faculty members via email.

Keywords: Attendance Report, AI, Face Recognition, Haar-Cascade classifier, Local Binary Pattern Histogram

Abstract: This project presents an IoT-powered Smart Dustbin system designed to automate waste management and improve cleanliness in urban areas. It uses a combination of ultrasonic, infrared (IR), and metallic proximity sensors, all connected to a NodeMCU microcontroller, to monitor how full the bin is and to help sort different types of waste. The system can detect and separate metallic from non-metallic waste, using servo motors to guide each type into its own mini compartment. To keep things efficient, the dustbin is also equipped with GSM and GPS modules. These allow it to send SMS alerts when the bin is full and share its real-time location, helping city workers collect waste on time and plan better collection routes. All sensor data is sent to a web dashboard built with Python Flask, giving authorities the ability to monitor everything remotely. By combining intelligent waste segregation with location tracking, this system offers a smarter, more scalable solution for managing waste. It reduces the need for manual handling, encourages cleaner surroundings, and supports broader smart city goals. Overall, it’s a step forward in creating a cleaner, more efficient, and sustainable urban environment.

Keywords: IoT, Smart Dustbin, Waste Segregation, NodeMCU, GSM, GPS, Python Flask, Proximity Sensor, Servo Motor, Real-Time Monitoring

Abstract: Enhanced spatial transformations are advanced techniques designed to improve the accuracy and quality of translating medical images from one form to another. These transformations help ensure that important details and shapes in the images are preserved during the translation process. This is crucial in medical imaging because doctors rely on accurate images for diagnosing and treating patients. By using enhanced spatial transformations, we can create translated images that closely resemble the original ones, making it easier for doctors to understand and analyze them. This approach combines spatial adjustments (like changing the position of image parts) and intensity adjustments (like altering brightness and contrast) to achieve more realistic and accurate results. The method has been tested successfully in various medical tasks, such as predicting future brain scans and visualizing changes in stroke- affected areas. Overall, enhanced spatial transformations significantly improve the quality of medical image translations, aiding in better clinical decisions and patient care.

Keywords: Spatial Intensity transform, image-to-image translation, Histogram Equalization, Generative Adversarial Network, Sharpening Filters, Noise Reduction, Lack of Ground Truth.

Abstract: The prediction of heart disease remains a critical challenge in healthcare, necessitating advanced computational methods to enhance diagnostic accuracy and patient outcomes. This study proposes a hybrid machine learning model integrating Convolutional Neural Networks (CNN) and extreme Gradient Boosting (XG-Boost) to improve heart disease prediction. The CNN component excels in automatically extracting complex features from diverse input data, including medical records, wearable device readings, and genomic information. These extracted features are then fed into the XG-Boost model, known for its robust classification capabilities, to accurately predict the presence or absence of heart disease.

Keywords: Hybrid machine learning, (CNN), (XG-BOOST), Data preprocessing, Performance Evaluation, Accuracy, Precision, Data privacy, Scalability, Gradient, boosting.

Abstract: This study uses deep learning techniques to predict the academic performance of postgraduate (PG) students. By analyzing data such as grades, attendance, and online activity, we trained models like Deep Neural Networks (DNN) and Recurrent Neural Networks (RNN). Results show DNN achieving approximately 89% accuracy, making it an effective tool for early intervention.

Keywords: Deep Learning, DNN, RNN, Data Mining.

Abstract: Massive amounts of time-series data have been produced as a result of the spread of Internet of Things (IoT) devices, which offers both opportunities and difficulties for analysis in highly dynamic and resource-constrained environments. In this study, methods for unsupervised anomaly detection and clustering in IoT time-series data based on deep learning (DL) are investigated. The performance of network analysis systems is severely hampered by important issues like noise, high dimensionality, and irregular sampling, which must be addressed. While noise can obscure subtle anomalies and result in high false positive rates, irregularity breaks temporal coherence, making it challenging to consistently identify patterns over time. High dimensionality has a detrimental effect on clustering accuracy and model interpretability because it raises computational complexity and may dilute important signals. These difficulties make it more difficult to deploy DL models in real time, particularly on resource-constrained edge devices. To address these problems, this project suggests a systematic strategy that combines algorithm development, theoretical modeling, and practical validation. The main objective is to improve the state of time-series analysis for IoT in order to facilitate anomaly detection, predictive maintenance, and more precise and effective monitoring. The results of this study have the potential to greatly improve the intelligence and dependability of IoT systems in a variety of fields by bridging the gap between theoretical innovation and real-world application. Index Terms: Long Short-Term Memory (LSTM), network analysis, anomaly detection, clustering, deep learning, time series data, Internet of Things (IoT), and high-dimensional data.

Abstract: This research paper presents an innovative approach to railway safety through the development of an Arduino Nano-based train accident prevention system. The system utilizes ultrasonic sensors for real-time obstacle detection on railway tracks, coupled with immediate alert mechanisms including audible buzzers and visual LED indicators. A distinctive feature of this implementation is the integration with Processing IDE software, which provides a graphical interface displaying real-time obstacle detection data for train operators. The prototype demonstrates effectiveness within a 1-meter detection range, offering a cost-effective solution (₹4,110) compared to conventional railway safety systems. The paper comprehensively covers the system design, implementation challenges, test results, and proposes future enhancements including IoT integration and machine learning applications for improved reliability under various environmental conditions.

Keywords: Railway Safety, Arduino Nano, Ultrasonic Sensor, Real-time Monitoring, Obstacle Detection, Embedded Systems

Abstract: The rapid incorporation of IoT technologies into most sectors of our day-to-day life, the health sector inclusive, has given room to the harnessing and analysis of data related to patients. However, the aged risk death through the worldwide problem of aging that has been burdensome in the recent past. Many IoT devices are designed to monitor, track, and record the actions of the elderly to reduce these hazards. In this regard, the presented paper develops novel dependable cloud-based remote system patient monitoring framework for IoT health detection. Most distinguished part of this research is that we rarely find a framework in the literature that is developed with a basis on a real-time system by taking into consideration heartbeat (BPM), blood oxygen (SpO2), and body temperature at once. Implementation and testing of this real-time system is divided into six distinctly separated phases for developing both hardware and software. In order to validate the performance of the proposed system, the data are collected from BOT-IoT datasets. The outcome enhances patient satisfaction, safe data transmission, and healthcare results as it shows that the proposed framework is more efficient than the compared protocols in terms of the decision time, which is 16.3 seconds for 46 features with an accuracy of 100%.

Keywords: anomaly detection; cloud computing; health monitoring system; healthcare IoT architecture real-time monitoring; secure data transmission

Abstract: Software defect prediction plays an important part in perfecting software quality and it help to reducing time and cost for software testing. Machine literacy focuses on the development of computer programs that can educate themselves to grow and change when exposed to new data. The capability of a machine to ameliorate its performance grounded on former results. Machine literacy improves effectiveness of mortal literacy, discover new effects or structure that's unknown to humans and find important information in a document. For that purpose, different machine literacy ways are used to remove the gratuitous, incorrect data from the dataset. Software defect prediction is seen as a largely important capability when planning a software design and much lesser trouble is demanded to break this complex problem using a software criteria and disfigurement dataset. Metrics are the relationship between the numerical value and it applied on the software thus it's used for prognosticating disfigurement. The primary thing of this check paper is to understand the being ways for prognosticating software disfigurement.

Keywords: Software Defect Prediction, Software Metrics, Machine Learning Techniques.

Abstract: Plant diseases threaten global food security, especially in high-demand crops like tomatoes and potatoes. Early and accurate detection is vital to minimize yield loss and maintain produce quality. Traditional methods, such as manual inspection, are often slow, error-prone, and require expert knowledge. With advancements in AI and computer vision, automated systems now enable faster and more accurate disease identification. This project uses the YOLO v11 algorithm, an advanced real-time object detection model, to detect diseases in tomato and potato plants. YOLO v11 improves feature extraction, detection precision, and localization, even under changing lighting and noisy backgrounds. By training on a diverse dataset of healthy and diseased plant images, the system can accurately differentiate between infections. The enhanced accuracy reduces false positives and negatives, ensuring more reliable detection. Early identification allows farmers to apply timely treatments, reducing pesticide use and preventing crop losses. Overall, this AI-powered system boosts agricultural productivity and promotes sustainable farming.

Keywords: YOLO v11, plant disease detection, tomato, potato, AI, computer vision, real-time object detection, crop management, agricultural sustainability, early intervention, precision agriculture.

Abstract: Ensuring the originality and appropriateness of titles is crucial in academic research, project submissions, and business naming. The Title Uniqueness Verification System leverages Natural Language Processing (NLP) to provide an automated solution for title validation. Developed as a Python Flask-based web application, the system allows users to register, log in, and submit titles through an intuitive interface. Upon submission, the system employs the Cosine Similarity algorithm to compare the submitted title with a dataset of existing titles. If the similarity score exceeds a predefined threshold, the title is rejected, preventing duplication and potential plagiarism. Additionally, the system integrates a keyword filtering mechanism to identify and reject titles containing disallowed or restricted words, ensuring compliance with specific content standards. This real-time, automated verification method helps researchers, academic institutions, and organizations maintain originality and adhere to content guidelines, significantly reducing redundancy and enhancing the integrity of title submissions.

Keywords: Title Verification, Natural Language Processing (NLP), Cosine Similarity, Plagiarism Detection, Keyword Filtering, Flask Web Application, Title Originality, Automated Title Validation, Academic Integrity, Content Compliance.

Abstract: HeritageConnect is an AI-powered multilingual platform designed to make Tamil Nadu’s cultural and historical heritage more accessible to a global audience. Leveraging natural language processing, machine translation, and voice technologies, the system allows users to engage in real-time, interactive conversations with a virtual guide. Users can ask questions in Tamil, English, Hindi, French, or Spanish—via text or voice—and receive informative, context-aware responses about heritage sites, architecture, festivals, and local traditions. The platform integrates a Flask backend with MongoDB for managing user sessions and chat history, and uses models like DeepSeek and Hugging Face for intelligent language processing. With a mobile-responsive frontend, voice support through the Web Speech API, and robust user personalization, HeritageConnect redefines how cultural knowledge is delivered, making it immersive, inclusive, and multilingual.

Keywords: Heritage Tourism, Multilingual Chatbot, Natural Language Processing, Voice Interaction, Cultural AI

Abstract: Conventional management tools frequently fail to provide real-time responsiveness, effective resource allocation, and seamless cross-platform support in the context of large-scale project execution. A Cross-Platform Application for Major Project Management and Tracking, created with Flutter, is presented in this paper as a comprehensive solution that guarantees device-agnostic access on desktops, tablets, and smartphones. The system solves important issues with current platforms, such as platform dependency, disjointed team communication, and inefficient task monitoring. The suggested approach uses a hierarchical task delegation model in which project managers give leads tasks to complete and then divide those tasks into smaller, deadline-driven tasks. Team members are always informed thanks to Firebase's real-time updates and notifications, which cut down on misunderstandings and delays. Better tracking of project milestones and the identification of workflow bottlenecks are made possible by the application's sophisticated data visualizations and user-friendly interface. The system maximizes resource utilization, facilitates well-informed decision-making, and improves project workflow transparency through integrated analytics and stringent data security procedures. In order to establish this cross-platform tool as a scalable, effective, and safe substitute for traditional project management services in contemporary, fast-paced settings, this study examines its conception, deployment, and expected effects.

Keywords: Flutter Development, Firebase, Cloud, SQL lite, Operations Management, Real-Time Collaboration, Database, Resource Allocation, Team Communication.

Abstract: Read Stability Enhancement for Soft Error Low-Power 12T SRAM with Multi-Node Upset Recoverability of Aerospace application and result of advances in technology. If a radiation particle have been to have an impact on a sensitive node of a normal 6T SRAM cell, the files that had been saved in the smartphone would be flipped, which would cease end result in a single-event upset (SEU). Comparisons are made between SARP12T and exceptional these days disclosed soft-error-aware SRAM cells. This lets in for an assessment of the relative universal overall performance of SARP12T. In addition to these benefits, the proposed 12T SRAM cellular telephone having most diploma of find out about stability, it will in distinction to majority of the cutting-edge. All of these enhancements to the proposed mobile phone may additionally additionally be carried out by using the use of having a learn about latency that is truly marginally longer and via capacity of having a study and write electrical energy consumption that is marginally greater.65nm and 45nm CMOS Technology and proved the comparisons of area, prolong and electrical energy the utilization of Tanner EDA Tool. Index terms: Data Storage, Read Operation, Energy Consumption, High Power, Larger Area, Power Consumption, Current Source, Inverter, Amount Of Charge, Availability Of Devices, High Power Consumption, Dynamic Power, Minority Carrier, Remote Memory.

Abstract: Freelancing platforms are essential in providing linkages between clients and independent experts thanks its features like project matching, payment jobs and evaluation. This work is primarily concerned with designing an AI based freelancing platform, which is expected to improve scaling, customization, and overall efficiency of operations. In turn, the system aspires to use AI-based systems to automatically recommend projects, implement dynamic pricing, and map freelancers’ skills more accurately to client requirements. A comparative analysis of the above objectives illustrates the benefits of AI technologies in increasing user activity and improving the accuracy of service provision. The paper also provides an analysis of the problems related to the use of AI and the various possible future directions such as the use of blockchain technologies as a mean for transaction execution and predictive demand forecasting. Through surveys of freelancing platforms, many stakeholders see both promise and the threat in the deployment of AI solutions.

Keywords: Freelancing Platform, AI Integration, Skill Matching, Scalability, Predictive Analytics

Abstract: The project implements a local machine learning system that uses Python to predict cryptocurrency prices through lightweight operations. The system unites data science flexibility with regression-based learning models to enable users to execute complete offline predictive model training and evaluation and deployment without server-based or cloud-based computing needs. Through its CSV data-input capability the application provides strong data preprocessing and analysis as well as an interface to try several regression models including Linear Regression Decision Trees and Random Forests. The system incorporates visualization components together with evaluation metrics to boost interpretability and usability features. Users can access the system through a basic interface which provides easy avenues for adding real-time feeds and deep learning models in addition to current capabilities. The paper examines the main framework design and data processing systems while outlining potential upgrades to validate easy-to-implement offline prediction technologies for business forecasting.

Keywords: Cryptocurrency, Machine Learning, Regression Models, Offline Prediction, Financial Forecasting

Abstract: IoT devices seem like easy targets to attackers because manufacturers limit their computing capability and maintain insufficient security defenses. The present paper provides extensive analysis about machine learning techniques that enhance the security of IoT networks. ML operates in a dual capacity where systems need specific design to maintain defense security together with protection against potential attacks. The research evaluates multiple machine learning models active in real-time intrusion detection systems and explains their weak points along with present-day IoT cybersecurity threats analysis. The study describes vital barriers alongside anticipated advancements that will lead to the development of secure intelligent IoT networks.

Keywords: IoT, Machine Learning, Cybersecurity, Anomaly Detection, Threat Prediction, Supervised Learning, Unsupervised Learning, Adversarial Attacks

Abstract: The introduction of social media has revolutionized how we communicate, interact socially, and conduct our daily lives worldwide. For as much as there are gains such as connectivity and access to information from social media, its possible negative contributions to mental health have been a concern. Several studies suggest that overuse of social media is associated with increased anxiety, depression, sleep issues, and poor concentration. Most of these claims have not been tested through critical examination of data, however. The purpose of this research is to investigate the link between social media usage and key mental health indicators through a quantitative, data-based methodology. This research includes both descriptive and prescriptive type of analysis which means that we have tried to offer some possible solutions based on the problems from the data. Our findings reveal significant associations between social media engagement and measures of mental health, with wide variation by gender. These results highlight the complex relationship that is present between virtual interaction and mental health. The results offer significant insights to policymakers, mental health professionals, and social media users, thereby indicating the necessity for responsible digital interaction and heightened mental health awareness. The findings of this study contribute to the knowledge of the multifaceted effect of online activity on psychological well-being. These findings have important implications for digital wellbeing policy, mental health practice, and responsible social media usage.

Keywords: Social Media, Mental Health, Data Analysis, Descriptive Analysis, Linear Regression

Abstract: This study introduces an innovative Intrusion Detection System (IDS) that leverages the combined strengths of Principal Component Analysis (PCA) and the Random Forest machine learning algorithm. The primary goal of this approach is to efficiently identify and classify network intrusions while minimizing data noise and enhancing computational performance. The proposed framework employs PCA to reduce the dimensionality of input data and utilizes a Random Forest classifier to accurately identify threats and malicious activities.The performance of the model was evaluated using the NSL-KDD dataset, a widely recognized benchmark for IDS research. The results demonstrate that integrating PCA and Random Forest creates a robust and efficient IDS capable of adapting to evolving cyber threats. The study also explores the system's implementation details, potential for integration with existing security infrastructure, and scalability for real-time applications. Future directions include exploring the use of deep learning models and unsupervised anomaly detection techniques to further advance intrusion detection capabilities.  

Keywords: Intrusion Detection System, Machine Learning, PCA, Random Forest, Network Security, Cybersecurity.

Abstract: In this project, a complete indoor positioning system is proposed, which utilizes Bluetooth Low Energy (BLE) beacon with iBeacon technology for location-based services in the buildings. The proliferation of tracking devices (smartphones with GPS embedded) equipped with certain low-power sensors such as accelerometers has transformed many aspects of human lives, also creating new opportunities. GPS technology works great for outdoor positioning but falters indoors due to satellite signal restrictions. The purpose of the project is to use iBeacon (one workflow, with more location power) technology and solve the problems of indoor positioning in a creative way. Low Energy Beacons (BLE) that come with iBeacon satisfy such characteristic and represent a built-in cross-platform technology for Android and iOS devices in the long run.The iBeacon technology offers a range of significant benefits that make it a valuable tool for various sectors, including retail, event management, and education. Its advantages include cost-effective hardware, lower energy consumption, independence from internet connectivity, and the capability to send notifications in the background. These features enhance communication and improve user experiences in indoor environments.Recent advancements in iBeacon projects have focused on integrating both X and Y coordinate predictions into a unified model. This approach employs a valid time series data split for training and testing, utilizing Convolutional Neural Networks (CNNs) to analyze sensor data spatially. A novel method transforms sensor readings into an image-like format, allowing CNNs to effectively capture spatial relationships. Additionally, unsupervised pretraining with autoencoders is leveraged to utilize unlabeled data, which can minimize the need for manual measurements in real-world settings.Initially, a Multilayer Perceptron (MLP) was used for position prediction, establishing a foundational understanding of how sensor inputs relate to coordinates. The transition to CNNs enhances spatial comprehension by treating sensor data as images, thereby improving generalizability across varying environments.

Keywords: Bluetooth Low Energy (BLE),iBeacon Technology,Multilayer Perceptron (MLP),Convolutional Neural Network (CNN), Received Signal Strength Indicator (RSSI).

Abstract: VisionSpeak ,an Android-based mobile application that enhances real-world awareness through intelligent object detection and text recognition. Using a smartphone camera, the app identifies objects and extracts printed or handwritten text in real time. Recognized information is instantly converted into speech using a Text-to-Speech (TTS) engine, allowing users to receive voice-based feedback without needing to look at the screen. The app integrates deep learning models like MobileNet and YOLO for efficient object detection and uses the Tesseract OCR engine for text recognition. Designed with accessibility in mind, it supports voice commands, offline functionality, and a user-friendly interface. VisionSpeak is particularly useful for individuals with visual impairments, travelers, and those seeking hands-free interaction. Its seamless performance across diverse environments makes it a versatile tool for daily assistance.

Keywords: Object Detection, Text Recognition, Android Application, Text-to-Speech (TTS), Assistive Technology.

Abstract: As cyber threats continue to rise, individuals and businesses face growing risks from phishing attacks. This project introduces an AI-driven Phishing Detection and Awareness Platform, designed to provide a comprehensive and interactive approach to phishing awareness, detection, and prevention. Developed as a web-based solution, the platform integrates phishing detection (AI), password breach checking, email analysis, URL scanning, and user awareness training. The phishing detection module analyzes emails and URLs using fine- tuned transformer models, while the breach checker leverages the HaveIBeenPwned API. quizzes and interactive simulations enhance user training, ensuring engaging and practical learning. Additionally, the platform includes real-time alerts, detailed reports, and secure data storage to strengthen cybersecurity measures. By offering an all-in-one, accessible, and AI-powered security solution, this project empowers users to recognize, prevent, and respond to phishing threats more effectively.

Keywords: AI-powered phishing detection, Phishing simulation, cybersecurity training, Real-time threat alerts, AI- driven email analysis, Automated phishing awareness, Secure password management, URL threat analysis, Cybersecurity education platform, AI-powered risk assessment, Interactive phishing scenarios, Data-driven phishing insights, AI in cybersecurity awareness, Smart anti-phishing solutions.

Abstract: A global user base consisting of millions groups together for internet access. User information along with data tampering and application intrusions are easy to achieve because of these weaknesses in the system. A proposed security system implements hybrid protection features which combine data hiding with three-stage graphical passwords and file split and merges techniques to protect users. The AES algorithm serves as our proposed approach for securing the daily key owner's private information. The proposed method works to stop intrusions along with delivering complete user privacy protection for data. Our project contains 3 stages of graphical passwords for secure authentication implementation. Our proposed security system implements three stages of authentication which defend both users and stops unauthorized intruders from logging in. The project implements authentication security through traditional login with username and password and also includes color sequence verification followed by graphical image matching as the final authentication stage. Human beings tend to recall images more easily than written text according to psychological research findings. The data owner file gets encrypted by AES encryption while split and merge implementation is applied. The file of the data owner becomes three separate parts that get stored safely in cloud storage to protect against unauthorized access of cloud service providers.

Keywords: Graphical passwords, AES encryption, Hybrid protection, Secure authentication, Cloud storage.

Abstract: The increasing sophistication of contemporary energy systems requires effective and secure Demand Side Management (DSM) solutions. Conventional centralized solutions tend to be inadequate owing to privacy issues, trust shortcomings, and communication hindrances. This paper advocates a blockchain-enabled Machine-to-Machine (M2M) communication paradigm for DSM in smart grids. Smart contracts and distributed ledger concepts are utilized to facilitate automated transactions between generators and power management systems for security, transparency, and decentralization. A 34-node microgrid simulation confirms the model's feasibility and effectiveness. Outcomes show decreased line overloads, self-sustained energy trading, and better load balancing by smart contract-based interactions.

Keywords: Blockchain, Demand Side Management, Smart Grid, M2M Communication, Smart Contracts, Distributed Energy

Abstract: In educational institutions, attendance monitoring is a crucial task that ensures students' regular participation in academic activities. Traditional manual methods are time-consuming and prone to errors, while biometric methods such as fingerprint scanning often cause delays due to long queues. This paper presents an efficient attendance management system using face recognition technology. The system captures student images through a camera, detects and recognizes faces using advanced machine learning techniques, and automatically records attendance. The methodology includes image preprocessing, feature extraction, and classification using the Principal Component Analysis (PCA) and Eigenface approach. Experimental results demonstrate high accuracy and efficiency compared to conventional methods.

Keywords: Principle Component Analysis, Convolutional Neural Networks, Facial Recognition, Image Acquisation, Feature Extraction.

Abstract: Cervical cancer ranks fourth globally among cancers that affect women, after lung, colorectal, and breast cancers. The WHO reports that there are over 4 lakh cervical cancer deaths and 6 lakh new cases annually which are highly concerning numbers. Cervical cancer is usually preventable and has significantly greater survival chances when detected early. In high-income countries with robust screening and vaccination programs, it is a rare disease. But the disease kills women in low- and middle-income countries, where there are scarcities of resources and hence it becomes a severe and irreversible illness. Treatment options include surgery, chemotherapy, and radiation therapy, either alone or in combination. This review paper discusses stages, complications, causes, and treatment plans for cervical cancer.

Keywords: Cervical cancer, Stages, Risk factors, Complications

Abstract: Accurate prediction of insurance payouts for car damage is essential for fair and efficient claim settlements in the insurance industry. This project introduces an innovative approach that leverages Generative Adversarial Networks (GANs) and deep learning techniques to estimate insurance amounts based on the severity of car damage. The system employs a GAN framework, where the generator creates synthetic images of damaged cars with varying severity levels, and the discriminator enhances the model's ability to recognize intricate damage patterns. These synthetic images are used to augment the training dataset, improving the model's performance. Features extracted from the images, combined with structured data such as car make, model, and accident details, are used to predict the insurance payout. This AI-driven method enhances prediction accuracy, reduces reliance on large labeled datasets, and improves generalization to new and complex damage scenarios. Automating the assessment process increases efficiency, reduces fraud, and ensures faster and more consistent claim processing.

Keywords: Insurance Amount Prediction, Car Damage Assessment, Generative Adversarial Networks (GANs),Deep Learning, Damage Severity Classification.

Abstract: In this digital world the advanced technology are grown up very fast and effective in the way of transforming the digitalized system. The human resources also be converted into the advance version of the technology world. The human resources technology system (HRTS) and Applicant Tracking Systems (ATS) are the critical components of this transformation. We provide a thorough explanation of our study process that includes the sampling test, collecting data from company staffs, and analyzing it. Then the final result of the research report is highlighted significant changes in these practices providing deep insight into the digital transformation and to exploring how HR technology provides customized career development paths, learning, sourcing and organizational development. This article provides the proof of the easiest way of HR practices and all the HR roles are done in a single formatted units by using the HR technology system. Research in the field of digital HR transformation has gained significant momentum due to the influence of business transformation driven by digital technology. The research on this topic remains relatively scattered and dispersed.

Keywords: Data Analysis, Human resources technology System, Applicant Tracking Systems, Digital HR Tools, HR Automation

Abstract: In modern industrial and construction environments, maintaining accurate employee attendance and enforcing strict safety compliance are vital components of efficient workforce management. This project presents an AI-driven, real-time attendance and safety monitoring system that leverages deep learning technologies to address these critical needs. The system integrates facial recognition using a high-precision deep learning model from the face recognition library and helmet detection using the YOLOv8 object detection algorithm. It ensures that attendance is marked only when an employee is both properly identified and wearing the required safety helmet, thereby promoting safety standards while eliminating identity fraud or proxy attendance. Captured data such as employee name, helmet status, and timestamp is stored securely in Firebase Firestore, providing real-time synchronization and robust cloud-based data management. An intelligent alert mechanism is also embedded into the system, which triggers a notification when unauthorized individuals or non-compliant workers are detected, enhancing on-site security and proactive incident response. A Flutter-based mobile application complements the system by providing real-time access to attendance and safety compliance records, offering a user-friendly interface for administrators and supervisors to monitor workforce activities. This intelligent framework not only automates routine attendance tasks but also supports scalable safety enforcement, contributing to a safer and more accountable working environment. By combining artificial intelligence, cloud computing, and real-time monitoring, the system paves the way for smarter workforce governance in safety-critical sectors.

Keywords: Face Recognition, YOLO, Helmet Detection, Firebase, Workplace Safety, Deep Learning, Real-Time AI.

Keywords: Soft Tissues Tumors, Diagnosis and Classification, Soft Tissue Tumor Management, and Machine Learning

Abstract: The PashuRaksak: IoT-Driven Automated Livestock Rescue System aims to address the critical need for protecting livestock from fire emergencies caused by accidents such as firecrackers, electric short circuits, and other fire outbreaks in rural areas. The system uses IoT technologies to detect fire in real time and automatically initiate a rescue protocol. This includes opening gates, releasing livestock, and notifying emergency responders such as firefighters. By minimizing human intervention and accelerating response times, the system enhances animal welfare, reduces losses, and supports the broader goal of creating smart, disaster-resilient agricultural communities.

Keywords: IoT, Livestock Safety, Fire Detection, Automated Rescue System, Smart Farming.

Abstract: The current ATM authentication method through PINs exposes users to vulnerabilities such as stolen PINs and cloned cards in traditional systems. This project introduces a Face ATM System that improves safety through deep learning facial identification and mobile authentication instead of the current PIN-based systems. Users receive Face Verification Links through their mobile phones to establish secure account access after CNNs verify their faces. The system delivers security alerts in real-time while keeping banks notified about each transaction to detect problematic behaviour. The system, developed with Python, Flask, OpenCV, and MySQL, presents a security-focused and fraud-resistant method that enhances the security profile and user experience of ATM transactions.

Keywords: Deep Learning, CNN, Biometric Authentication, Mobile Verification, AI-driven Authentication.

Abstract: Stress among students is a growing concern, impacting academic performance, mental health, and overall well-being. Traditional methods for detecting stress such as self-assessment surveys and physiological measurements are often invasive, subjective, or impractical in real-time educational settings. In recent years, image-based facial expression recognition has emerged as a non-intrusive and efficient approach to detect stress levels using advancements in computer vision and machine learning. This literature survey presents an overview of recent techniques and models developed for stress-level detection through facial expressions, emphasizing their application in student populations. We analyze various datasets, image preprocessing methods, facial emotion recognition algorithms, and stress classification frameworks. The study also identifies current limitations and highlights research gaps to support the development of an improved, real-time, image-based stress detection system for educational institutions.

Abstract: Alzheimer's disease is a progressive neurodegenerative condition that significantly affects cognitive abilities, resulting in memory impairment, confusion, and challenges in carrying out everyday tasks. A significant hurdle for individuals with Alzheimer’s is their difficulty in recognizing family members and caregivers, which can lead to emotional distress and increased dependency. Moreover, the failure to remember to take prescribed medications exacerbates their health issues. To tackle these challenges, this research introduces a laptop-based assistive system that combines deep learning facial recognition technology with a medication reminder feature, aimed at improving the quality of life for those with Alzheimer’sThe system utilizes a laptop's webcam to capture real-time facial images, which are then analyzed by a deep learning model to identify familiar faces. Upon recognizing a known individual, the system announces their name audibly, aiding the patient in recalling and recognizing their loved ones. Furthermore, it includes a medication reminder function that notifies patients at specific times to promote adherence to their prescribed treatment regimen. This solution is designed to be standalone and user-friendly, negating the need for additional IoT devices, thus making it suitable for home environments.This research outlines the design, implementation, and assessment of the assistive system, focusing on its accuracy, usability, and effects on patient care. The facial recognition component employs Convolutional Neural Networks (CNNs) for accurate identification, while the medication reminder utilizes a structured scheduling system with audio-visual alerts. Experimental findings demonstrate that the system effectively aids patients in recognizing individuals and following their medication schedules.By merging AI-powered facial recognition with intelligent reminders, this assistive technology seeks to promote patient independence, alleviate caregiver stress, and enhance overall well-being. The study also considers potential future enhancements, including improved emotion detection capabilities.

Abstract: This research introduces an AI-driven communication platform designed to bridge the communication gap between hearing and deaf individuals. The system employs Temporal Convolutional Networks (TCNs) for precise recognition of Indian Sign Language (ISL) gestures. To handle spoken language input, it incorporates a Speech Recognition and Synthesis Module (SRSM) that utilizes Hidden Markov Models (HMMs) to transcribe speech into text. A 3D avatar module subsequently interprets the transcribed speech into ISL visual gestures, allowing for seamless real-time interaction. The gesture recognition model, trained on the MNIST ISL dataset, achieved a high accuracy rate of 98.5%, ensuring dependable performance in both gesture-to-text and speech-to-sign translation tasks. Designed for inclusivity, the system caters to deaf, mute, and non-signing users. Additionally, a user-friendly web interface enhances accessibility and ease of use across platforms.

Keywords: Sign Language, Temporal Convolutional Network, Speech Recognition, Indian Sign Language, 3D Avatar, Accessibility, Deep Learning, Inclusive Communication.

Abstract: The road safety is an important aspect in the present scenario. The project aims to improve the road safety by using machine and deep learning to monitor and classify driver behaviour in real time. It identifies ten types of activities of driver including- safe driving, texting, phone usage, drinking and more. The system uses advances CNNs, transfer learning models like VGG16 and ResNet50, and YOLOv8 for object detection. It also includes a drowsiness detection module to alert drivers showing signs of fatigue. The project uses the state farm distracted driver detection dataset for training and evaluation, and flask-based web app for real-time monitoring and alerts. Performance is measured using Accuracy, Precision, Recall, and F1-score, showing high effectiveness in enhancing driver awareness and reducing accidents. This system is suitable for modern vehicle safety and fleet management solutions. The drowsiness module is also integrated to alert the driver feeling drowsy and improve the safety. It utilizes the standard dataset of open and closed eyes for training and detects the drowsy behaviour in real time.

Keywords: ML, Road safety, VGG16, ResNet50, YOLOv8, CNN, Real-time monitoring.

Abstract: This paper demonstrates about Visually impaired students face significant challenges in accessing educational materials in traditional learning environments. The proposed system is an Android application designed to enhance accessibility through voice assistance. The application provides functionalities such as document scanning and text-to-audio conversion using Optical Character Recognition (OCR), enabling students to listen to printed and digital content. Additionally, features like voice-activated assistance, bookmarking key sections, voice recording, and note-taking empower users to engage in independent learning. By integrating advanced speech recognition and web search capabilities, the system enhances accessibility and promotes a seamless educational experience for visually impaired students.

Keywords: Visually Impaired Education ,Optical Character recognition (OCR), Voice Assistant, Text to Speech, Accessible Learning Technologies.

Abstract: The Real-Time Alert System for Crime Area Mapping is designed to enhance public safety through advanced technologies like geospatial mapping and real-time data analysis. This system provides instant notifications to citizens, law enforcement, and security agencies about crime incidents within specific locations, enabling swift preventive measures and improved situational awareness. By utilizing Geographic Information System (GIS) technology, crime data is visualized on interactive maps, allowing authorities to allocate resources effectively and patrol high-risk areas more efficiently. Citizens receive real-time alerts through mobile notifications, helping them take necessary precautions. Additionally, predictive analytics integrated into the system aids in crime trend analysis, supporting law enforcement in proactive decision-making.This technology-driven approach not only enhances security but also contributes to urban planning and policy development by offering insights into crime patterns. Ultimately, the system serves as a valuable tool for improving public safety, optimizing law enforcement responses, and fostering a more secure urban environment.

Keywords: Real-time alert system, Crime area mapping, Public safety, Geographic information system (GIS).

Abstract: This paper presents the design, implementation, and evaluation of an intelligent Learning Management System (LMS) that integrates advanced features such as semantic search using a knowledge graph (Neo4j), fine-grained access control for content and role-based authentication. Our LMS incorporates automated tagging, multilingual document indexing, and a permission workflow for secure content delivery. In a comparative analysis with Moodle, our experimental evaluations demonstrate improvements in retrieval accuracy, latency in access control, and overall user satisfaction. These results highlight the potential of our LMS to serve as a next-generation educational platform that supports fairness and enhanced usability.

Keywords: Learning Management System, Moodle, Knowledge Graph, Semantic Search, Access Control, Educational Technology.

Abstract: This paper introduces the development of a room allocation system using a custom algorithm to optimize room and staff assignments for examinations, taking into account factors such as room capacity, staff availability, and fairness. The system utilizes React.js for the frontend, Firebase for authentication, and Node.js for backend processing. Our analysis reveals that the system significantly improves allocation efficiency over traditional methods, optimizing space utilization and reducing scheduling conflicts. Key contributions include a scalable allocation algorithm and an automated scheduling framework.

Keywords: Room allocation, scheduling, custom algorithm, React.js, Firebase, Node.js, real-time updates, optimization, automation, room capacity, staff availability

Abstract: Parental control applications are essential for promoting healthy digital habits among children. This paper introduces Kid Care, a parental control system for Android mobiles and Android TVs that enforces strict screen time limits. Once the allowed time expires, the screen is automatically disabled and can only be reactivated with a secure passcode, ensuring parental supervision. The system includes uninstallation protection to prevent unauthorized removal. Kid Care is evaluated for its usability, security, and performance, addressing challenges like power efficiency and bypass prevention. Future improvements involve integrating AI-based monitoring and adaptive usage analytics. The study highlights the growing importance of reliable digital parenting solutions in today's connected environment.

Keywords: Parental Control, Screen Time Management, Child Safety, Digital Parenting, Android TV, App Security, Uninstallation Protection, Multi-Device Synchronization

Abstract: New technologies needs data to be refined and narrowed down for further processing on it. Big data concept fixes this problem of dealing with large amount of data by performing algorithms which are much easier than the traditional methods which are more complex time consuming, costly, and requires high amount of space. As we are working on data refining, we need to work with raw data which requires high amount of space physically obtaining this space is highly expensive as we require such hardware and software. Also, the refined data or the end product of such hefty data is huge so its storage also requires huge space. Cloud computing provides a platform where the problem of storage is solved. In our paper we present the correlation of both Big Data and Cloud Computing together. We can work on and store huge data with ease. Providing a user and pocket friendly platform. We will discuss on topic of analysis of Cloud based big data in Microsoft Azure, Amazon Web Service, Google Cloud using.

Keywords: Big Data, Cloud Computing, Microsoft Azure, Google Cloud, Amazon Web Service

Abstract: Real analysis is an area of analysis that studies concepts such as sequences and their limits, continuity, differentiation, integration and sequences of functions. By definition, real analysis focuses on the real numbers, often including positive and negative infinity to form the extended real line.

Abstract: Plant leaf diseases are a major contributor to reduced agricultural productivity and economic loss worldwide. Traditional detection methods rely on manual inspection, which is time-consuming, inconsistent, and inadequate for large-scale implementation. This paper proposes an IoT and Deep Learning-based Smart Plant Disease Detection System designed to overcome these limitations. The system employs Convolutional Neural Networks (CNNs) in MATLAB to accurately classify plant diseases from leaf images. To enhance detection and enable precision agriculture, an IoT framework incorporating NodeMCU, DHT11 temperature and humidity sensors, and soil moisture sensors is used for real-time environmental monitoring. Sensor data is transmitted to the ThingSpeak cloud platform, where it is analyzed to facilitate intelligent irrigation control and early disease alerts. By integrating deep learning with environmental sensing, this system provides a scalable and automated approach for early disease detection and optimized resource usage in agriculture.

Keywords: Plant Disease Detection, IoT in Agriculture, Convolutional Neural Network (CNN), Smart Irrigation, ThingSpeak, Environmental Monitoring, MATLAB, Precision Farming.

Abstract: This project proposes an AI-powered web application intended for effective event inspection and automatic report generation. Developed with Streamlit, the system allows the uploading of event-related images from which essential metadata like date and geolocation are parsed using EXIF data. The application uses natural language processing (NLP) methods, in this case, sentiment analysis through TextBlob, to assess the emotional tone of event descriptions. Via a secure admin login, event information—name, organizer, description, and location—can be entered and tracked. In-review events are inspected and digitally signed by authorized individuals via an upload or real-time drawing canvas. Upon approval, the system creates a professional PDF report with inlined images, metadata, sentiment summary, and signatures via ReportLab, and securely stores it on Cloudinary. This automation is not only onerous to documentation but also accurate, standardized, and easily accessible. The combination of AI and cloud services converts conventional event reporting into an intelligent, quick, and dependable process that is appropriate for institutional and organizational settings. Keywords- AI-driven application, event inspection, automated report generation, Streamlit, image metadata extraction, EXIF data, sentiment analysis, natural language processing (NLP), TextBlob, geolocation, digital signature, PDF report generation, ReportLab, Cloudinary, web-based system, administrative approval, event management automation, user authentication, institutional documentation.

Abstract: With the present era of computer technology, the ability to successfully transform numerous inputs into editable and usable text is increasingly essential. This "Multi-Mode Text Converter" project aims to bridge the gap between digital and non-digital inputs and digital documentation through an easy, accessible, and multi-purpose web-based program. Streamlit is used to develop the app, which comprises two primary functionalities: Image to Text conversion using Optical Character Recognition (OCR) and Voice to Text transcription via Speech Recognition. The Image to Text (OCR) feature permits users to import printed or cursive images and derive text content from them by employing Tesseract OCR, supported with Tamil and English languages. Preprocessing processes such as thresholding and grayscale conversion are utilized to enhance text recognition accuracy and improve image quality. For English texts, TextBlob is also employed by the app for automatic spell checking and correction for the generation of quality text output. Extracted or edited text may be exported in a.txt format for convenience to use in future purposes. The Voice to Text module utilizes the Google Speech Recognition API to transcribe live voice input captured from a microphone. It is possible to choose between English and Tamil speech recognition, and users enjoy regional inclusivity as well as support for multilinguality. Transcribed text is also storable for documentation and archival purposes. One of the key features of the application is the Text-to- Speech (TTS) feature, powered by gTTS (Google Text-to-Speech), through which users can listen to the recorded or typed text in their chosen language. For improved user experience, the application uses an appealing graphical user interface with a customized background and adaptive layout. By integrating image processing, natural language editing, speech recognition, and voice synthesis, the Multi-Mode Text Converter is an end-to-end system for digital text creation and extraction. It also has its future applications in education, accessibility software, digital archiving, and simple-to- use data entry systems.. Keywords- Optical Character Recognition (OCR), Automatic Speech Recognition (ASR), Image to Text, Voice to Text, Tesseract OCR, TextBlob, gTTS, Speech Recognition API, Streamlit, multi-language support, text-to-speech, image preprocessing, digital text conversion, handwritten text recognition, audio transcription, user interface, AI integration.

Abstract: As computers play an increasingly vital role in human life and daily activities across various domains, humans have leveraged their intelligence and creativity to use computers in natural and effective ways. Hence, a reliable method for recognizing handwritten digits is essential. Handwritten Digit Recognition (HDR) can offer a clear benefit in this aspect. Deep Learning (DL) has been a powerful tool for solving various problems with high accuracy in recent years

Abstract: Secret Image Sharing (SIS) is a secure method of disseminating an image by breaking it into n shadow images where k number is required to reconstruct the image. The existing techniques of SIS suffer from a security loophole and inefficient storage and are susceptible to tampering. Thus, this paper proposes a Blockchain-based Secure and Optimized SIS (BC-SOSIS) scheme to overcome those issues. The scheme enables decentralized storage to resist tampering, tightly coupled with smart contracts to achieve authentication, along with efficient encryption that ultimately improves its security and performance. Furthermore, the security analysis as well as experiments carried out against BC-SOSIS validate it as a scalable and reliable solution for the secure digital communication.

Keywords: Blockchain, Secure Image Sharing, Smart Contracts, Decentralized Storage, IPFS, Multi-Secret Image Sharing.

Abstract: DDoS attacks pose a significant threat to Software-Defined Networking (SDN) environments, often overwhelming traditional security mechanisms. This work is primarily concerned with designing an AI-driven DDoS detection and mitigation system, which is expected to improve scalability, adaptability, and overall efficiency of network security operations. The system aspires to use AI-based models, including Multi-Armed Bandit, Random Forest, and Online Gradient Boosting, to dynamically detect anomalies, classify attack traffic, and implement intelligent mitigation strategies in real time. A comparative analysis of these models illustrates the benefits of AI technologies in enhancing detection accuracy, reducing false positives, and optimizing network performance. The paper also provides an analysis of the challenges associated with AI-based intrusion prevention and explores various future directions, such as the use of federated learning for collaborative threat intelligence sharing. Through studies on AI-based cyber security solutions, many researchers recognize both the potential and challenges in the deployment of real-time, adaptive DDoS mitigation strategies.

Keywords: DDoS Mitigation, AI-Driven Security, SDN Protection, Multi-Armed Bandit, Online Gradient Boosting, Anomaly Detection, Threat Intelligence.

Abstract: The fast-evolving strides of artificial intelligence, specifically using generative adversarial networks (GANs), have ushered in the era of deepfakes—artificial media capable of replicating human faces, voices, and actions with comparative ease. Although the technology has revolutionary and positive applications across the domains of filmmaking and accessibility, it equally bears colossal risks if used for cyber fraud. Deepfakes are being used more and more by cybercriminals for impersonation, identity theft, business email compromise (BEC), and various other types of deception. Impersonation of CEOs over video calls, audio message fakes to approve illegal fund transfers, and evading biometric security controls using synthetic faces and voices with a hyper-realistic appearance are now achievable by fraudsters. The existing reality of deepfakes as a tool for cybercrime is examined in this paper. It discusses actual events where deepfakes were utilized to take advantage of, deceive, or financially exploit individuals and groups. Moreover, it has a detailed description of the detection methods created to help counter this emerging threat. These include some of them being passive detection methods like artifact and frequency analysis, deep learning classifiers, and biological signal detection, and others being active detection methods like liveness checks, watermarking, and blockchain-based content verification. Despite concerted efforts, the race between deepfake generation and detection remains on an upward trajectory. Attackers continue to adapt to remain undetected, and conventional forensic mechanisms become less effective with time. The paper concludes on a note highlighting the importance of hybrid detection systems, robust regulatory frameworks, and global cooperation to enable ethical and secure use of AI-generated content.

Keywords: Artificial Intelligence, Deepfakes, Cyber Fraud, Generative Adversarial Networks (GANs), Deepfake Detection, Identity Theft, Liveness Detection, Biometric Security, AI Forensics, Cybersecurity.

Abstract: While the international digital universe continues to expand more and more interconnected, network security can no longer be overstated. Internet use continues to boom as cloud computing and remote work are now the standards, yet networks have also become top targets for all manner of cyberattack from malware and ransomware to clever phishing attacks, and APTs. As companies become increasingly dependent on networked infrastructure in day-to-day operations, protecting the integrity, security, and availability of such networks is no longer an added luxury but now a minimum concern. It is the purpose of this paper to discuss the pillars of network security in terms of their theoretical foundations like the basic pillars of confidentiality, integrity, availability, authentication, and non-repudiation. It also addresses common forms of threats and vulnerabilities that have the potential to compromise network security, such as Denial of Service (DoS) attacks, sniffing, spoofing, and social engineering. To counter these threats, a plethora of security technologies and mechanisms have been developed and implemented. They range from firewalls, intrusion detection and prevention systems (IDPS), encryption methods, Virtual Private Networks (VPNs), and secure authentication protocols. On top of these, sound security policies, employee training initiatives, and regular audits form the people and procedural aspect of an integrated security infrastructure. The article also highlights the importance of upcoming trends such as Zero Trust Architecture, artificial intelligence-driven cybersecurity, and quantum cryptography, which reflect the direction of network security innovation over the next few years. By assembling these essential but core concepts, this document is a student manual, IT professional, and business manual to learn and develop their network security function. The purpose is to put into perspective the reality that cybersecurity is not an event, but a continuous process which demands watchfulness, rapidity, and foresight so that they can fight prospective threats.

Keywords: Network Security, Cybersecurity, Confidentiality, Integrity, Availability, Authentication, Encryption, Firewall, Intrusion Detection, VPN, Phishing, Malware, Zero Trust Architecture, Quantum Cryptography, AI in Security

Abstract: The development of Internet of Things (IoT) devices in critical infrastructures—namely, electric grids, health systems, transport systems, and industrial control networks—has ushered in monumental advantages in terms of automation, efficiency, and data-informed decision-making. Yet, the digital revolution has also been linked to heightened vulnerability to cybersecurity attacks in the form of data breaches, distributed denial-of-service (DDoS) attacks, spoofing, and unauthorized access. Decentralized security paradigms are now falling short to cope with the distributed, heterogeneous, and resource-limited nature of IoT networks. Here, the convergence of Blockchain and Artificial Intelligence (AI) technologies presents an end-to-end and promising solution to securing IoT. Blockchain's distributed, secure ledger guarantees data integrity, secure device authentication, transparent logging, and tamper-proof device communication. Policy enforcement and secure access control capabilities can be automated via smart contracts. AI completes this framework with context-aware analytics features such as anomaly detection, real-time threat anticipation, behaviour monitoring, and automated incident response. This work investigates a Blockchain-AI hybrid architecture specifically to secure IoT environments in critical infrastructures. We introduce a multi-layered architecture that welcomes edge computing, federated learning, and smart contracts to provide an efficient, scalable, and secure security model. The system seeks to detect advanced cyber-attacks, automate response activities, and provide secure peer-to-peer communication in a distributed device network. Along with describing the technicalities of this merged model, the paper also tackles significant challenges—latency, energy efficiency, and scalability—and comes across areas for upcoming research like lightweight consensus algorithms and privacy-preserving AI models. The real-world applications of these in smart grid, healthcare, and industrial automation are evaluated to suggest the real-world application and efficacy of the proposed solution. This research is intended to assist in the formulation of future-proof cybersecurity frameworks using the potential of Blockchain and AI to establish smart, autonomous, and decentralized IoT security infrastructures.

Keywords: Internet of Things (IoT), Blockchain, Artificial Intelligence (AI), Critical Infrastructure, Cybersecurity, Smart Contracts, Anomaly Detection, Decentralized Security, Federated Learning, Edge Computing