Archives

• 

  Data Trustworthiness in Mobile Crowd Sensing

  Vol. 1 No. 03 (2024)

  Abstract

  The project, "Data Trustworthiness in Mobile Crowd Sensing," aims to address the critical issue of ensuring the reliability and authenticity of data collected through mobile crowd sensing applications. In the rapidly evolving landscape of sensor-equipped smartphones and ubiquitous connectivity, leveraging the collective intelligence of a crowd for data acquisition has become increasingly popular. However, the inherent challenges of ensuring the trustworthiness of data gathered from diverse sources pose significant obstacles.

  This project focuses on developing robust mechanisms and algorithms to validate and authenticate data in the context of mobile crowd sensing. The research encompasses the design and implementation of stringent data collection protocols, authentication measures, and quality control mechanisms to filter out inaccurate or fraudulent data points. The goal is to enhance the overall reliability of information collected from various contributors.

  In addition to technical aspects, the project emphasizes the importance of creating a transparent and collaborative environment. Privacy-preserving techniques and clear communication regarding data usage policies are integral components to foster trust among contributors. By addressing these aspects, the project aims to establish a framework that ensures the anonymity and privacy of participants while building a foundation of trust in the mobile crowd sensing ecosystem.

  Ultimately, the outcomes of this project are expected to contribute significantly to the advancement of reliable data collection practices in mobile crowd sensing applications, fostering innovation in areas such as environmental monitoring, urban planning, and healthcare.

  Index Terms

  Mobile Crowd Sensing, Data Trustworthiness, Data Authentication, Data Validation, Reliability, Authenticity, Sensor-equipped Smartphones, Ubiquitous Connectivity, Collective Intelligence, Data Collection Protocols, Quality Control Mechanisms, Fraudulent Data, Privacy-preserving Techniques, Data Usage Policies, Transparency, Collaboration, Privacy, Anonymity, Environmental Monitoring, Urban Planning, Healthcare.

• 

  Innovative Dual Authentication Protocols for Cloud Data Storage and Sharing

  Vol. 2 No. 01 (2025)

  Abstract

  As cloud computing continues to revolutionize the way data is stored and shared, the security of sensitive information becomes a paramount concern. This BTech project delves into the design and implementation of a robust dual access control framework for cloud-based data storage and sharing. The proposed system aims to enhance the security posture of cloud environments by incorporating a two-tier authentication mechanism.

  The dual access control system combines traditional authentication methods, such as usernames and passwords, with an additional layer of security, such as biometrics or multi-factor authentication. This two-pronged approach ensures that only authorized users can access and share data stored in the cloud, minimizing the risk of unauthorized access and potential data breaches.

  The project will involve the development of a prototype system, including the integration of authentication mechanisms and the establishment of secure data sharing protocols. Through rigorous testing and evaluation, the effectiveness of the dual access control system in safeguarding sensitive information will be assessed.

  The outcomes of this project are expected to contribute to the advancement of cloud security practices, providing a valuable solution for organizations and individuals seeking enhanced protection for their data in cloud environments. The project aligns with the growing demand for innovative approaches to address the evolving challenges of securing information in the era of cloud computing.

  Index Terms

  Cloud computing, Data storage, Data sharing, Security, Dual access control, Authentication mechanisms, Two-tier authentication, Biometrics, Multi-factor authentication, Prototype development, Secure data sharing protocols, Testing and evaluation, Sensitive information, Data breaches, Cloud security practices, Innovative approaches, Organizational security, Individual security, Evolution of challenges.

• 

  Integrated E-Commerce Platform for Agriculture: Enhancing Supply Chain Efficiency and Farmer Empowerment

  Vol. 2 No. 04 (2025)

  Abstract

  The " Integrated E-Commerce Platform for Agriculture Enhancing Supply Chain Efficiency and Farmer Empowerment" is a comprehensive technological solution designed to address the evolving needs of the agricultural sector. This project aims to create a user-friendly online platform that facilitates seamless transactions and collaboration between farmers, agribusinesses, and suppliers. The portal integrates various modules, including inventory management, order tracking, and payment processing, to streamline the agricultural supply chain.

  The primary focus is on providing farmers with easy access to a diverse range of agricultural products and services, such as seeds, fertilizers, and equipment. The portal also serves as an information hub, offering valuable insights on best practices, market trends, and innovative farming techniques. By fostering a digital marketplace, the project aims to empower farmers, enhance their decision-making processes, and contribute to the overall efficiency and sustainability of the agricultural ecosystem.

  Through this Integrated E-Commerce Portal, the project envisions creating a connected and collaborative environment for stakeholders in the agricultural value chain. This initiative not only supports farmers in making informed choices but also promotes transparency and efficiency in agricultural transactions. Overall, the project aims to leverage technology to uplift the agricultural community, fostering growth, sustainability, and improved productivity in the sector.

  Index terms

  Integrated E-Commerce Portal, Agriculture, Technological solution, User-friendly platform, Seamless transactions, Collaboration, Farmers, Agribusinesses, Suppliers, Inventory management, Order tracking, Payment processing, Agricultural supply chain, Agricultural products, Services, Seeds, Fertilizers, Equipment, Information hub, Best practices, Market trends, Farming techniques, Digital marketplace, Empowerment, Decision-making processes, Efficiency, Sustainability, Stakeholders, Value chain, Transparency, Growth, Productivity.

• 

  Building a Network Traffic Analyzer for Security Monitoring

  Vol. 1 No. 05 (2024)

  Abstract

  The project titled "Building a Network Traffic Analyzer for Security Monitoring" addresses the critical need for robust cybersecurity measures in today's interconnected digital landscape. With the ever-increasing threat of cyber-attacks, organizations require advanced tools to monitor and safeguard their network infrastructure. This project aims to develop a sophisticated network traffic analyzer that provides real-time monitoring and analysis of network activities for enhanced security.

  The proposed system focuses on capturing and analyzing network traffic using state-of-the-art technologies, ensuring a comprehensive view of the data traversing the network. The project incorporates machine learning algorithms to detect anomalies and suspicious patterns, enabling the identification of potential security threats.

  The user interface is designed with a user-friendly dashboard, featuring intuitive visualizations and tools to empower security professionals in quickly interpreting the data. The system includes alerting mechanisms to promptly notify administrators of any detected anomalies, enabling swift response to potential security incidents.

  This project not only emphasizes the technical aspects of building a powerful network traffic analyzer but also addresses the importance of cybersecurity in safeguarding sensitive information. The implementation of this system contributes to the enhancement of network security protocols, providing organizations with a valuable tool to proactively monitor and mitigate potential risks.

  Keywords

  Network Traffic Analyzer, Security Monitoring, Cybersecurity, Real-time Monitoring, Network Activities, Machine Learning Algorithms, Anomaly Detection, User Interface, Dashboard Visualization, Alerting Mechanisms, Network Security Protocols, Risk Mitigation, Digital Landscape, Data Analysis, Threat Detection

• 

  Enhanced Weather Forecasting: Integrating Firefly Optimization with Deep Recurrent Neural Networks

  Vol. 2 No. 03 (2025)

  Abstract

  The project "Weather Prediction Using Firefly Optimization and Deep Recurrent Neural Networks (DRNN)" explores the integration of two powerful techniques, Firefly optimization and DRNN, to enhance weather forecasting accuracy. Weather prediction is crucial for various industries and sectors, including agriculture, transportation, and disaster management. However, the inherent complexity and dynamic nature of weather systems pose significant challenges to accurate forecasting. Traditional forecasting methods often struggle to capture intricate temporal patterns and dependencies present in meteorological data.

  In this project, the team proposes a novel approach that combines the strengths of Firefly optimization, a metaheuristic algorithm inspired by the flashing behavior of fireflies, with DRNN, a type of neural network tailored for sequential data analysis. Firefly optimization is employed to optimize the parameters of the DRNN model, facilitating efficient training and enhancing its predictive capabilities. By leveraging Firefly optimization's ability to effectively explore the solution space and DRNN's capacity to capture temporal dependencies, the integrated approach aims to improve the accuracy of weather predictions.

  The project involves the implementation and experimentation of the proposed methodology using real-world weather datasets. Performance evaluations will be conducted to assess the effectiveness of the combined approach in comparison to traditional forecasting methods. The outcomes of this project are expected to contribute to the advancement of weather prediction techniques, offering potential benefits in terms of improved forecasting accuracy and reliability. Additionally, the project provides valuable insights for researchers and practitioners in the field of meteorology and related domains.

   

  Index Terms

  Weather Prediction, Firefly Optimization, Deep Recurrent Neural Networks (DRNN), Forecasting Accuracy, Meteorological Data, Traditional Forecasting Methods, Temporal Patterns, Sequential Data Analysis, Solution Space Exploration, Parameter Optimization, Performance Evaluation, Real-World Datasets, Advancements in Weather Prediction, Meteorology Research, Disaster Management

• 

  ADVANCED AI-DRIVEN SYSTEM FOR VECHICLE CLASSIFICATION AND AUTOMATED NUMBER PLATE RECOGNITION

  Vol. 1 No. 1 (2024)

  ABSTRACT

  This paper presents an advanced AI-driven system designed for vehicle classification and automated numberplate recognition (ANPR). Leveraging state-of-the-art machine learning algorithms, the system offers robust performance in diverse environmental conditions. The core components include convolutional neural networks (CNNs) for image processing and optical character recognition (OCR) for number plate detection. The proposed solution demonstrates high accuracy and speed, making it suitable for real-time applications in traffic management and law enforcement. This innovative approach not only improves the efficiency and accuracy of vehicle identification but also reduces the need for manual intervention, paving the way for smarter, more autonomous traffic systems.The AI-driven system is built on a scalable architecture that can be easily integrated into existing traffic management infrastructures. This flexibility ensures that the system can be adapted to various urban and rural settings, addressing the unique challenges posed by different environments. The incorporation of machine learning models enables continuous improvement of the system’s performance over time, as it learns from new data and adapts to changing conditions. The paper discusses the technical details of the system, including the data preprocessing, model training, and deployment processes.

  INDEX TERMS

  Vehicle Classification, Automated Number Plate Recognition, Convolutional Neural Networks, Optical Character Recognition, Machine Learning, Traffic Management, Law Enforcement, Real-Time Processing, Image Processing, Deep Learning.

• 

  Enhancing Network Security through Intrusion Detection Systems

  Vol. 1 No. 03 (2024)

  Abstract

  The project titled "Enhancing Network Security through Intrusion Detection Systems" aims to explore and implement advanced techniques to strengthen the security posture of computer networks. In today's digital landscape, where cyber threats are ever-evolving, the role of Intrusion Detection Systems (IDS) becomes crucial in detecting and mitigating potential security breaches.

  The project focuses on the comprehensive understanding and deployment of both signature-based and anomaly-based IDS approaches. It delves into the development of a robust system capable of monitoring network traffic, identifying known attack patterns, and detecting deviations from normal behavior. By combining these methods, the project aims to provide a more effective defense against a wide range of cyber threats.

  Furthermore, the project incorporates the integration of machine learning algorithms within the IDS framework. This addition allows the system to learn and adapt to emerging threats, thereby improving its ability to detect previously unknown and sophisticated attacks. The implementation of machine learning contributes to a dynamic and intelligent intrusion detection mechanism, reducing false positives and enhancing overall accuracy.

  The outcomes of this project will not only contribute to the academic understanding of network security but will also provide practical insights into implementing advanced intrusion detection techniques. Ultimately, the project seeks to empower organizations with a more resilient defense against cyber threats, ensuring the confidentiality, integrity, and availability of their networked systems.

• 

  Zero-Shot Multilingual Sentiment Analysis Using Transformer-Based Models

  Vol. 2 No. 02 (2025)

  Abstract

  This project aims to explore the feasibility and effectiveness of zero-shot multilingual sentiment analysis using transformer-based models. Traditional sentiment analysis techniques often rely on language-specific models trained on large corpora of labeled data, making them impractical for analyzing sentiments across multiple languages. In contrast, transformer models, such as BERT and GPT, have shown promising results in natural language understanding tasks by leveraging large-scale pre-training and fine-tuning on specific tasks. This project proposes to extend the capabilities of transformer models to perform sentiment analysis across various languages without requiring language-specific training data. The project will involve pre-training a transformer model on multilingual text data and fine-tuning it on sentiment analysis tasks using transfer learning techniques. The effectiveness of the proposed approach will be evaluated on standard benchmark datasets in multiple languages, measuring the accuracy and robustness of sentiment predictions. The outcomes of this project have the potential to significantly enhance the applicability of sentiment analysis tools in multilingual settings, catering to diverse linguistic communities and enabling broader cross-cultural sentiment analysis applications.

  Index terms

  Zero-shot sentiment analysis, Multilingual sentiment analysis ,Transformer-based models ,BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pre-trained Transformer), Natural language understanding, Transfer learning, Pre-training, Fine-tuning, Sentiment analysis tasks, Benchmark datasets, Accuracy measurement, Robustness assessment, Cross-cultural sentiment analysis, Linguistic diversity, Applicability enhancement, Language-specific models, Large corpora, Feasibility study, Effectiveness evaluation.

• 

  Development of an LSTM-Based System for Real-Time Toxic Comment Classification in Online Platforms

  Vol. 1 No. 05 (2024)

  Abstract

  The project aims to develop an efficient and accurate system for classifying toxic comments in online platforms using Long Short-Term Memory (LSTM) networks. With the exponential growth of online content, ensuring a safe and respectful environment for users is of paramount importance. Toxic comments, characterized by offensive, abusive, or harmful language, pose a significant challenge for content moderation.

  This project employs LSTM, a type of recurrent neural network, known for its ability to capture sequential dependencies in data. The LSTM model will be trained on a labeled dataset comprising both toxic and non-toxic comments. During training, the model learns to recognize patterns and contextual cues associated with toxic language, enabling it to make predictions on unseen comments.

  The implementation involves preprocessing textual data, constructing an LSTM architecture, and fine-tuning the model to achieve optimal performance. The trained LSTM model will be integrated into an interactive platform, providing real-time classification of comments. The project will also explore techniques for model evaluation, addressing challenges such as false positives and false negatives.

  The successful implementation of this project can significantly contribute to enhancing online content moderation, fostering healthier digital communication spaces, and mitigating the impact of toxic behavior. Additionally, the project offers valuable insights into the application of deep learning techniques, specifically LSTM, in addressing real-world social challenges related to online content.

  Index Terms

  Toxic comments, Online platforms, Long Short-Term Memory (LSTM) networks, Content moderation, Recurrent neural network, Sequential dependencies, Labeled dataset, Textual data preprocessing, Model fine-tuning, Real-time classification, Model evaluation, False positives, False negatives, Digital communication spaces, Deep learning techniques.

• 

  Digital Campus Tour Realistic 3D Walkthrough with Interactive Features

  Vol. 2 No. 03 (2025)

  Abstract

  This project aims to develop a Virtual Campus Walkthrough application utilizing the three.js library, offering an immersive and interactive experience for users to explore a digital representation of a campus environment. The application will allow users to navigate through the virtual campus using keyboard or mouse inputs, providing a simulated experience akin to physically walking through the real campus. Leveraging the capabilities of three.js, the application will render high-quality 3D graphics and textures, ensuring a visually compelling and realistic environment. Users will have the opportunity to interact with various elements within the virtual campus, such as buildings, landmarks, and informational points of interest. The project will focus on creating a user-friendly interface, optimizing performance for smooth navigation, and integrating interactive features to enhance the overall experience. Ultimately, the Virtual Campus Walkthrough aims to provide a valuable tool for prospective students, faculty, and visitors to explore and familiarize themselves with the campus remotely.

  Index Terms

  Virtual Campus Walkthrough, three.js library, Immersive Experience, Interactive Exploration, Digital Representation, Campus Environment, Navigation Controls, 3D Graphics, Textures Rendering, Realistic Environment, User Interaction, User-Friendly Interface, Performance Optimization, Interactive Features, Prospective Students, Faculty, Visitors, Remote Exploration.

• 

  PERSONALIZED SKINCARE PRODUCTS RECOMMENDATION SYSTEM USING ML AND DL

  Vol. 1 No. 1 (2024)

  Abstract The personalized skincare product recommendation system leverages machine learning (ML) and deep learning (DL) techniques to provide tailored product suggestions based on individual skin profiles. The system utilizes a combination of user input, skin condition analysis, and product data to deliver recommendations that address specific skincare needs. By integrating advanced ML algorithms and DL models, the system aims to enhance user experience, improve skincare outcomes, and drive engagement with personalized recommendations. This paper details the design, implementation, and evaluation of the system, demonstrating its efficacy in delivering accurate and relevant skincare product recommendations.Index TermsPersonalized Skincare, Product Recommendation, Machine Learning, Deep Learning, Skin Analysis, Recommendation Systems, User Profiling, AI in Beauty, Skincare Data, PredictiveModeling, Natural Language Processing (NLP), Convolutional Neural Networks (CNNs), Recommender Systems, User Experience.1. IntroductionPersonalized skincare is an emerging field where individual skin characteristics and needs drive the selection of appropriate products. Traditional skincare product recommendations are often generic, lacking personalization and not accounting for unique skin conditions. Advances in machine learning (ML) and deep learning (DL) offer a transformative approach to building sophisticated recommendation systems that can analyze user data, such as skin type, concerns, and preferences, to suggest tailored skincare solutions.Machine learning algorithms enable the processing of large datasets, extracting patterns, and making predictions based on historical user data and product efficacy. Deep learning models, particularly those using convolutional neural networks (CNNs) and recurrent neural networks (RNNs), enhance the system's capability to understand complex relationships between user profiles and product features. This integration allows for a more precise and individualized approach to skincare recommendations.The need for personalized skincare solutions is driven by increasing consumer demand for products that address specific skin concerns such as acne, aging, or sensitivity. Traditional approaches to skincare often rely on broad categorizations that may not fully address individual needs. By incorporating ML and DL into the recommendation system, the proposed solution aims to bridge this gap, providing users with personalized product suggestions that are more likely to meet their specific requirements.This paper explores the development of a personalized skincare recommendation system, detailing the methodologies employed, thedesign of the recommendation algorithms, and the evaluation of system performance. The integration of ML and DL techniques into the recommendation process offers a novel approach to enhancing user satisfaction and optimizing skincare outcomes.      
• 

  CCTV-Based Safety Gear Detection System for Enhancing Workplace Safety in Industrial Environments

  Vol. 1 No. 04 (2024)

  The implementation of CCTV surveillance for safety gear detection in industrial environments is becoming increasingly imperative to enhance workplace safety standards. This project aims to develop a robust system utilizing CCTV cameras to detect the presence or absence of essential safety gear worn by employees, such as helmets, goggles, and gloves. Through image processing techniques and machine learning algorithms, the system will analyze real-time video feeds to identify instances of non-compliance with safety regulations. The project will involve the design and development of algorithms for object detection and classification, as well as the integration of these algorithms into a user-friendly interface for industrial monitoring purposes. The anticipated outcome is a reliable CCTV surveillance system capable of proactively identifying safety gear violations, thereby mitigating potential hazards and promoting a safer work environment in industrial settings.

  Index terms

  CCTV surveillance, Safety gear detection, Industrial environments, Workplace safety standards, Image processing techniques, Machine learning algorithms, Object detection, Classification algorithms, Real-time video feeds, Non-compliance detection, Safety regulations, User-friendly interface, Industrial monitoring, Safety gear violations, Hazard mitigation, Proactive identification, Work environment safety, Robust system design.

   

• 

  Intelligent Fake Profile Detection for Instagram Using Machine Learning

  Vol. 2 No. 02 (2025)

  Abstract:

  The project entails the development of a Python-based system for detecting fake Instagram profiles using machine learning models and image processing techniques. The script employs libraries such as TensorFlow, Keras, Scikit-Learn, Pandas, Matplotlib, Seaborn, and OpenCV, facilitating extensive data handling, statistical analysis, visualization, and machine learning model training. The process begins by importing relevant libraries and datasets, followed by preliminary data exploration and visualization to understand the features and distributions within the data. The training dataset is preprocessed and scaled before being fed into a neural network model structured with multiple dense layers and dropout for regularization, optimized using Adam optimizer for categorical cross-entropy loss. The model's performance is evaluated using metrics like accuracy and loss progression during training/validation phases. Further, the script incorporates image processing techniques to analyze profile pictures using Haar Cascade classifiers for face detection, which contributes to determining the authenticity of profiles. Additionally, OCR (Optical Character Recognition) via Easy OCR is applied to extract textual data from images, providing a comprehensive understanding of profile components such as username, posts, followers, and other descriptive elements. Overall, this project integrates multiple computational techniques to create an automated tool that aids in identifying and analyzing fake profiles on Instagram, showcasing the practical application of AI and machine learning in social media management.

  Keywords- Instagram, Fake Profile Detection, Python, Machine Learning, TensorFlow, Keras, Neural Networks, Data Visualization, Image Processing, OpenCV, Haar Cascade, OCR, EasyOCR, Pandas, Matplotlib, Seaborn, Scikit-Learn, Statistical Analysis, Data Preprocessing, Model Evaluation, Text Extraction, Profile Analysis.

• 

  Brain Tumor Detection Using Convolutional Neural Network

  Vol. 1 No. 002 (2024)

  Abstract— Brain Tumor segmentation is one of the most crucial and arduous tasks in the terrain of medical image processing as a human-assisted manual classification can result in inaccurate prediction and diagnosis. Moreover, it is an aggravating task when there is a large amount of data present to be assisted. Brain tumors have high diversity in appearance and there is a similarity between tumor and normal tissues and thus the extraction of tumor regions from images becomes unyielding. In this paper, we proposed a method to extract brain tumor from 2D Magnetic Resonance brain Images (MRI) by Fuzzy C-Means clustering algorithm which was followed by traditional classifiers and convolutional neural network. The experimental study was carried on a real-time dataset with diverse tumor sizes, locations, shapes, and different image intensities. In traditional classifier part, we applied six traditional classifiers namely Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Multilayer Perceptron (MLP), Logistic Regression, Naïve Bayes and Random Forest which was implemented in scikit-learn. Afterward, we moved on to Convolutional Neural Network (CNN) which is implemented using Keras and Tensorflow because it yields to a better performance than the traditional ones. In our work, CNN gained an accuracy of 97.87%, which is very compelling. The main aim of this paper is to distinguish between normal and abnormal pixels, based on texture based and statistical based features.

   

  Keywords— CNN, FCM, Medical Image, segmentation, SVM

• 

  A Hybrid Approach for Securing Digital Images Using Encryption and Steganography

  Vol. 2 No. 01 (2025)

  Abstract

  In today's digital age, ensuring the security and confidentiality of image data is paramount, particularly with the proliferation of online communication and storage platforms. This project proposes a hybrid approach for enhancing image security by synergistically integrating encryption and steganography techniques. The combination of these two methods aims to fortify the protection of digital images against unauthorized access and tampering. Encryption ensures the confidentiality of image content by converting it into an unintelligible form using robust cryptographic algorithms. Concurrently, steganography conceals the encrypted data within the image itself, embedding it imperceptibly into the pixels or metadata. The synergy of encryption and steganography not only safeguards the integrity of image data but also adds an additional layer of concealment, making it arduous for adversaries to detect and decipher sensitive information. Through this project, the efficacy of the hybrid approach will be evaluated through experimentation and analysis, with the ultimate goal of providing a comprehensive solution for image security in various applications and contexts.

   

• 

  LSTM-Based Telugu Text Generation: Enhancing NLP for Content Creation and Language Preservation

  Vol. 2 No. 04 (2025)

  Abstract

  In this project, a Telugu text generator utilizing Long Short-Term Memory (LSTM) neural networks is developed to facilitate the automatic generation of coherent and contextually relevant Telugu text. The LSTM architecture, known for its effectiveness in handling sequential data, particularly in natural language processing tasks, serves as the foundation for this text generation system. The project aims to address the growing demand for computational tools capable of generating Telugu text, catering to various applications such as content creation, language learning, and cultural preservation. Through extensive training on a large dataset of Telugu text, the LSTM model learns the underlying patterns, semantics, and grammatical structures of the language, enabling it to generate text that closely resembles human-written Telugu content. The project's implementation involves data preprocessing, model training, and evaluation phases, with a focus on optimizing the LSTM network's performance in generating fluent and coherent Telugu text. The developed Telugu text generator provides a valuable resource for Telugu speakers, researchers, and developers interested in leveraging natural language processing techniques for Telugu language applications.

  Index Terms

  Telugu text generation, Long Short-Term Memory (LSTM) neural networks, Sequential data processing, Natural language processing (NLP), Computational linguistics, Text generation systems, Language modeling, Data preprocessing, Model training, Evaluation metrics, Grammatical structures, Semantic analysis, Cultural preservation, Language learning, Content creation, Telugu language applications, Fluency assessment, Coherence analysis, Text similarity, Resource optimization.

• 

  Human Activity Recognition

  Vol. 1 No. 02 (2024)

  Abstract

  The project on "Human Activity Recognition" explores the implementation of sensor fusion and machine learning techniques to accurately identify and classify human activities in various scenarios. This research aims to contribute to the development of intelligent systems capable of understanding and responding to human behavior. The project focuses on utilizing data from diverse sensors, including accelerometers and gyroscopes, to capture intricate patterns associated with activities such as walking, running, sitting, and other movements.

  The proposed system employs advanced machine learning algorithms for the analysis of sensor data, aiming to enhance the accuracy and efficiency of human activity recognition. By integrating multiple sensor modalities, the project aims to create a robust model capable of handling diverse environmental conditions and activity contexts. The research explores the

  application of deep learning methodologies to extract meaningful features and patterns from the sensor data, enabling real-time and context-aware activity recognition.

  The significance of this project lies in its potential applications across various domains, including healthcare, sports analysis, and smart home environments. The developed system could find applications in health monitoring, personalized fitness tracking, and creating adaptive environments that respond to human behavior. The project not only contributes to the academic understanding of human activity recognition but also provides practical implications for the development of smart and responsive technologies in our daily lives.

• 

  Zero-Shot Multilingual Sentiment Analysis Using Transformer-Based Models Exploring Feasibility and Effectiveness

  Vol. 1 No. 04 (2024)

  This project aims to explore the feasibility and effectiveness of zero-shot multilingual sentiment analysis using transformer-based models. Traditional sentiment analysis techniques often rely on language-specific models trained on large corpora of labeled data, making them impractical for analyzing sentiments across multiple languages. In contrast, transformer models, such as BERT and GPT, have shown promising results in natural language understanding tasks by leveraging large-scale pre-training and fine-tuning on specific tasks. This project proposes to extend the capabilities of transformer models to perform sentiment analysis across various languages without requiring language-specific training data. The project will involve pre-training a transformer model on multilingual text data and fine-tuning it on sentiment analysis tasks using transfer learning techniques. The effectiveness of the proposed approach will be evaluated on standard benchmark datasets in multiple languages, measuring the accuracy and robustness of sentiment predictions. The outcomes of this project have the potential to significantly enhance the applicability of sentiment analysis tools in multilingual settings, catering to diverse linguistic communities and enabling broader cross-cultural sentiment analysis applications.

• 

  AI-Driven Hybrid System for Personalized Movie Recommendations

  Vol. 2 No. 02 (2025)

  Abstract:

  This paper introduces a novel hybrid recommendation system that integrates content-based and collaborative filtering approaches using deep learning techniques to enhance movie recommendations. Our model merges the metadata of movies, including genres, cast, and crew from the MovieLens dataset with user ratings to construct a comprehensive feature set. We employ a Term Frequency-Inverse Document Frequency (TF-IDF) vectorizer to extract content-based features and utilize Singular Value Decomposition (SVD) to derive collaborative filtering features, thereby addressing both user preferences and item characteristics.We further enhance the model by concatenating these features into a unified representation, which is then processed through a deep neural network to predict movie ratings. The network architecture consists of multiple dense layers with dropout regularization to prevent overfitting, ensuring robustness in learning complex user-item interactions. We evaluate our model on a standard dataset, focusing on mean squared error (MSE) as the performance metric to assess accuracy.The results demonstrate the effectiveness of our hybrid approach in providing precise recommendations by leveraging both the semantic content of movies and the historical interactions of users, thereby outperforming traditional methods that rely on singular recommendation strategies. This research contributes to the recommendation system community by showcasing a scalable and efficient method to improve recommendation quality and user satisfaction in multimedia services.

  Keywords- Recommendation systems, Deep learning, Hybrid models, Collaborative filtering, Content-based filtering, MovieLens dataset, TF-IDF vectorization, Singular Value Decomposition, Neural networks, User-item interaction, Multimedia services, Movie recommendations, Personalization, Machine learning, Artificial intelligence.

• 

  AI-Powered Mobile Application for Visually Impaired People

  Vol. 1 No. 03 (2024)

  Abstract

  This paper introduces an innovative mobile application powered by Artificial Intelligence (AI) technology, designed to significantly improve the daily lives of visually impaired individuals. The application incorporates advanced features such as object recognition, color recognition, offline functionality, currency notes recognition, barcode reading, text reader. These features collectively empower users by providing real-time information about their surroundings, facilitating safe and independent, enabling secure financial transactions, and enhancing overall accessibility. The integration of AI algorithms ensures accurate and efficient performance, making the application a valuable tool for improving the quality of life and promoting inclusivity for visually impaired individuals. This research contributes to the field of assistive technology by showcasing the potential of AI-driven solutions in addressing accessibility challenges and fostering independence among diverse user groups.

  keywords

  Artificial Intelligence, Mobile Application, Accessibility, Assistive Technology, Color Recognition, Offline Functionality, Object Recognition, Currency Notes Recognition , Barcode Reader , Text Reader , Scene Recognition, Voice Based Navigation

• 

  Missing Child Identification System using Deep Learning and Multiclass SVM

  Vol. 1 No. 02 (2024)

  Abstract

  The "Missing Child Identification System using Deep Learning and Multiclass SVM" is a groundbreaking project designed to address the pressing issue of locating and identifying missing children. Leveraging advanced technologies in the realms of deep learning and machine learning, this project aims to create a robust system for facial recognition and classification.

  The deep learning component of the system utilizes state-of-the-art techniques to extract intricate facial features, generating comprehensive representations of each child. Simultaneously, a multiclass Support Vector Machine (SVM) is employed to classify and refine the identification process. The SVM acts as a classifier, distinguishing between different classes of facial features, thereby enhancing the accuracy of categorizing missing children.

  The integration of deep learning and multiclass SVM in this project facilitates a more effective and efficient means of matching facial characteristics with existing databases. The result is a powerful tool that streamlines the identification process, enabling authorities to quickly and accurately reunite missing children with their families.

  This project not only showcases the potential of cutting-edge technologies in addressing social issues but also underscores the significance of technology-driven solutions in humanitarian efforts. The "Missing Child Identification System" stands as a testament to the positive impact that technology can have on society, particularly in safeguarding the well-being and security of our most vulnerable population – our children.

  Index terms

  Missing Child Identification, Deep Learning, Multiclass SVM, Facial Recognition, Machine Learning, Social Impact, Humanitarian Technology, Child Safety, Database Matching, Technology Solutions, Vulnerable Populations, Reunification Efforts, Advanced Technologies, Facial Feature Extraction, Classifier Systems

• 

  Student Dropout Prediction using DeepFM Algorithm

  Vol. 2 No. 01 (2025)

  Abstract

  This project revolves around the development and implementation of a predictive model for anticipating student dropout utilizing the innovative DeepFM algorithm. DeepFM, a fusion of deep learning and factorization machines, proves to be a potent tool for unraveling complex patterns within diverse datasets. The project aims to leverage the algorithm's capability to analyze and interpret multifaceted features such as academic performance, attendance, and socio-economic factors. Through the integration of deep neural networks and collaborative filtering, DeepFM demonstrates proficiency in discerning intricate relationships, contributing to accurate student dropout predictions. The project's primary objectives include the exploration of DeepFM's suitability for handling sparse and high-dimensional student data, as well as its potential to offer proactive insights for educational institutions to address and mitigate dropout risks. The anticipated outcome is a robust predictive model that aids educational institutions in identifying potential dropout candidates, facilitating timely interventions and support mechanisms to enhance student retention.

  Index Terms

  Predictive modeling, Student dropout prediction, DeepFM algorithm, Deep learning, Factorization machines, Multifaceted features, Academic performance, Attendance, Socio-economic factors, Deep neural networks, Collaborative filtering, Sparse data handling, High-dimensional data, Dropout risk mitigation, educational institutions, Timely interventions, Support mechanisms, Student retention, Data analysis, Machine learning applications.

• 

  Few-Shot Learning for Efficient and Accurate Crop Disease Detection in Agriculture

  Vol. 2 No. 04 (2025)

  Abstract

  Crop diseases pose a significant threat to agricultural productivity, necessitating rapid and accurate detection methods. This project explores the application of N-shot/Few-shot learning techniques for crop disease classification and detection. The conventional challenge in agricultural datasets is the scarcity of labeled examples for various disease classes. Leveraging the principles of N-shot learning, the project aims to develop a model capable of accurately identifying and categorizing crop diseases with only a limited number of labeled instances per class.

  The project involves the implementation of a machine learning model that learns to generalize from a small dataset, thus overcoming the limitations of insufficient labeled samples. By adopting N-shot learning, the system becomes adept at recognizing patterns and features associated with different crop diseases, enabling it to make reliable predictions even in scenarios with sparse labeled data. The proposed solution holds promise for resource-constrained environments where obtaining extensive labeled datasets is challenging.

  Through the development and evaluation of the N-shot/Few-shot learning model, this project contributes to the advancement of efficient and cost-effective crop disease detection methods. The outcomes are expected to have implications for sustainable agriculture by providing farmers with a reliable tool for early disease identification, thereby aiding in timely and targeted interventions to safeguard crop yield and quality.

  Index terms

  Crop diseases, Agricultural productivity, N-shot learning, Few-shot learning, Crop disease classification, Disease detection methods, labeled datasets, Machine learning models, Generalization, Pattern recognition, Feature extraction, Sparse labeled data, Resource, constrained environments, Efficient detection methods, Cost-effective solutions, Early disease identification, Timely interventions, Sustainable agriculture, Crop yield, Crop quality.

• 

  Empowering Community Bonds through Resources and Sharing

  Vol. 1 No. 05 (2024)

  Abstract

  The project titled "Empowering Community Bonds through Resources and Sharing" seeks to explore and implement initiatives that foster a sense of unity and collaboration within a community. The project aims to identify and establish shared resources, such as community gardens, tool libraries, or communal spaces, to facilitate the exchange of tangible resources. Additionally, it delves into the exchange of ideas, experiences, and support systems, promoting a culture of inclusivity and understanding among community members.

  Through this project will investigate the impact of resource-sharing initiatives on community development, self-sufficiency, and resilience. The implementation of practical solutions, such as creating platforms for knowledge exchange and collaborative projects, will be a key focus. The project aims to contribute to the enhancement of community bonds by emphasizing the importance of collective well-being and interdependence.

  By conducting research, designing and implementing community-based interventions, and evaluating the outcomes to demonstrate the positive effects of empowering community bonds through resources and sharing. The findings and recommendations from this project can serve as a valuable guide for future community development initiatives, promoting a more cohesive and resilient society.

  Index terms

  Community empowerment, Resource-sharing initiatives, Community development, Self-sufficiency, Resilience, Knowledge exchange platforms, Collaborative projects, Collective well-being, Interdependence, Community-based interventions, Unity and collaboration, Inclusivity, Community bonds, Shared resources, Community gardens, Tool libraries, Communal spaces, Impact evaluation, Positive effects, Future community development initiatives.

• 

  AI-Driven Stock Market Forecasting: Leveraging News Sentiment and NLP for Predictive Analytics

  Vol. 2 No. 03 (2025)

  Abstract

  This project delves into the innovative realm of stock price prediction by leveraging the insights derived from news articles. In an era characterized by the abundance of financial information and the rise of natural language processing techniques, the project aims to develop a predictive model that harnesses the power of textual data to forecast stock price movements.

  The project involves the collection of relevant news articles from diverse sources, with a focus on financial news that could impact stock markets. Natural language processing algorithms are implemented to analyze the sentiment and content of these articles, extracting key features that may influence stock prices. The dataset is then utilized to train machine learning models, employing techniques such as sentiment analysis and time-series analysis.

  The predictive model's performance is evaluated using historical stock data, measuring its accuracy in forecasting price movements against actual market outcomes. The project also explores the impact of various external factors, such as market trends, economic indicators, and geopolitical events, on the accuracy of predictions.

  Through this project aims to contribute to the evolving field of financial technology by providing a comprehensive exploration of the feasibility and effectiveness of utilizing news articles for stock price prediction. The project's findings may offer valuable insights into the integration of natural language processing and machine learning techniques in financial analysis, paving the way for advancements in predictive modeling within the context of dynamic and unpredictable stock markets.

  Index terms

  Stock price prediction, News articles, Natural language processing (NLP), Financial information, Machine learning models, Sentiment analysis, Time-series analysis, Historical stock data, Market trends, Economic indicators, Geopolitical events, Financialtechnology (FinTech), Predictive modeling, Financial analysis, Integration of NLP and machine learning, Dynamic stock markets.