Published in 2025
D.Dhanalakshmi, K.Santhiya
Emotion discovery from sound and video information has developed as a pivotal inquire about region in human-computer interaction, healthcare, security, and excitement. The capacity to precisely recognize emotions from multimodal inputs such as discourse and facial expressions empowers more sympathetic and context-aware frameworks. This paper presents an diagram of strategies and strategies utilized in emotion discovery utilizing both sound and video information. Sound highlights, counting prosody, pitch, tone, and cadence, are analyzed to capture passionate prompts in discourse, whereas video information centers on facial expressions, look, and body dialect. Combining both modalities upgrades emotion acknowledgment precision, leveraging the complementary qualities of each methodology. Profound learning models, such as Convolutional Neural Systems (CNNs) for video and Repetitive Neural Systems (RNNs) for sound, are broadly utilized to capture transient designs and spatial highlights. The challenges of multimodal combination, information awkwardness, and real-world changeability are examined, in conjunction with promising arrangements and assessment measurements. This ponder points to contribute to the advancement of more modern emotion discovery frameworks that can be conveyed over differing applications, such as virtual collaborators, computerized reconnaissance, and mental wellbeing observing.
Multimodal Emotion Discovery, Facial Expression Acknowledgment, Sound Highlights (Pitch, Tone, Cadence), Video Highlights
L. Haripriya, S. Kavipriya, N. Geetha
Digital transactions have grown exponentially with the rise of online banking, e-commerce and financial technology. However, this growth has also led to an increase in fraudulent activities such as credit card fraud, phishing, account takeovers and cryptocurrency scams. Traditional rule-based fraud detection methods struggle to keep up with evolving fraud tactics, making advanced artificial intelligence (AI) and machine learning-based models essential for fraud prevention. This survey explores various types of digital transactions, common fraud detection techniques using data mining and recent research contributions in the field. The study on supervised, unsupervised and hybrid machine learning approaches discussing their advantages and limitations. Additionally, we highlight the key challenges in implementing effective fraud detection systems and examine future trends, including real-time analytics, blockchain integration, behavioural biometrics and explainable AI, to enhance the security of digital transactions. This will be helpful to the researchers for their future research direction in this area.
Digital transactions, fraud detection, credit card fraud, cryptocurrency scams, machine learning, data mining, supervised learning and unsupervised learning
P.Valarmathi, S.Gayathri
In the digital age, the overwhelming availability of books across various genres and formats makes it challenging for readers to discover literature that aligns with their preferences. This study presents a machine learning-powered book recommendation system that provides personalized suggestions to enhance user experience. The system integrates collaborative filtering and content-based filtering techniques to analyze user preferences and book characteristics, ensuring relevant recommendations. The dataset comprises user ratings, book descriptions, and metadata, which are processed to extract meaningful features. Machine learning algorithms, including k-Nearest Neighbors (k-NN) and matrix factorization techniques, are employed to train the model and identify patterns in user behavior. To ensure recommendation accuracy, the system's performance is evaluated using metrics such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Results indicate the effectiveness of the proposed approach in delivering tailored book suggestions, enhancing the reading experience. Furthermore, this work highlights future improvements, including the integration of natural language processing (NLP) for enhanced content evaluation and user engagement, showcasing the potential of AI-driven recommendation systems in the literary domain.
Personalized Suggestions,k-Nearest Neighbors,Matrix Factorization,Mean Absolute Error,Root Mean Squared Error, User Behavior Analysis,Feature Extraction,Natural Language Processing
R.Kavitha, S.Ruthra
Facial Expression Recognition (FER) is a significant task in human–computer interaction, affective computing, and behavior analysis. Conventional methods rely on handcrafted features, which often fail to generalize over variations in illumination, pose, and occlusion. Deep learning, particularly convolutional neural networks (CNNs) and transformer-based architectures, has dramatically advanced FER performance by automatically learning hierarchical features from raw image data. This paper explores recent developments in deep learning–based FER, including CNNs, recurrent neural networks (RNNs), attention mechanisms, and multimodal approaches. We discuss the impact of large-scale labeled datasets, data augmentation, and domain adaptation techniques on improving model robustness. Experimental results demonstrate that deep learning models outperform traditional methods, achieving state-of-the-art accuracy on benchmark datasets such as FER-2013, CK+, and RAF. Despite these advances, challenges remain in handling real-world variability, class imbalance, and interpretability. Future directions include leveraging self-supervised learning, few-shot learning, and hybrid architectures to further enhance FER performance and generalization.
Feeling Acknowledgment, Convolutional Neural Systems, Profound Learning, OpenCV Real-Time Preparing, Facial Expression Investigation, Include Extraction
S.Dhanalakshmi, S.Devika
The Beauty Hub Management System is a comprehensive and user-friendly software solution designed to streamline the operations of beauty salons and wellness centers. This system integrates key functionalities such as customer management, service scheduling, staff coordination, and inventory control into a centralized digital platform. A core feature of the system is its online booking facility, which allows clients to conveniently schedule appointments based on service availability and staff preferences. The booking module includes real-time calendar views, automated confirmations, and reminders via email or SMS, significantly reducing no-shows and manual coordination. Additionally, the system provides administrative tools for managing services, pricing, customer feedback, and analytics to enhance operational efficiency and customer satisfaction. By digitizing day-to-day activities, the Beauty Hub Management System aims to improve the overall client experience while empowering salon owners with better control and insights into their business.
Beauty Hub Management System, Customer Management, Service Scheduling, Inventory Control, Online Booking, real-time Calendar, Automated Reminders, Appointment scheduling
S.Kiruthika, M. Devi
Brain tumors require early and accurate detection for effective treatment. Traditional diagnostic techniques rely on manual assessment of MRI scans, which can be time-consuming and prone to errors. Artificial Intelligence (AI) and deep learning models, particularly Convolutional Neural Networks (CNNs), have demonstrated high accuracy in tumor detection but often lack interpretability. This study employs Explainable AI (XAI) techniques, integrating ResNet50 with Gradient-weighted Class Activation Mapping (Grad-CAM) to enhance transparency in brain tumor detection.The dataset used comprises MRI images labeled for tumor presence, which underwent preprocessing and augmentation for improved model performance. Our proposed approach achieved high detection accuracy while providing visual explanations, ensuring trust and usability in clinical settings. The integration of Grad-CAM enabled radiologists to interpret and validate AI-based diagnoses effectively, bridging the gap between deep learning-based automation and medical decision-making.This work contributes to AI-driven medical diagnostics by balancing accuracy with interpretability, promoting wider adoption in healthcare. Future directions include testing the model on larger, more diverse datasets and incorporating additional interpretability techniques to further enhance reliability and trustworthiness.
Explainable AI, Brain tumor detection, MRI images, Deep learning, Grad-CAM, ResNet50, Medical image analysis
S. Menaka, S.Kiruthika Sree
Cloud Data Centers (CDCs) are vital to today's digital infrastructure, supporting services like web browsing and media streaming. However, rising electricity consumption and maintenance costs pose significant challenges for service providers. This research introduces the Maintenance Energy Costs Data Center (MECDC) algorithm, which optimizes maintenance and energy expenses. By utilizing dynamic server state management—Active Mode and Sleep Mode—along with modeling maintenance costs and energy consumption patterns, MECDC achieves substantial savings. Evaluations in real-world scenarios show that MECDC outperforms traditional load-balancing and energy-saving strategies.
Cloud Data Centers (CDC), maintenance costs, electricity consumption, server sleep mode, virtual machine migration, MECDC algorithm, energy efficiency, cost optimization, cloud computing
V.Reshma, Dr.N.Jean Effil
This paper presents an enhanced Hotel Booking Management System incorporating artificial intelligence to deliver personalized food recommendations. The system employs a role-based architecture comprising Admin, User, and Owner modules, facilitating secure and structured access control. It automates the end-to-end process of user registration, hotel booking, food ordering, and feedback collection. One of the standout components is its AI-powered food recommendation engine based on collaborative filtering. By analyzing historical data and identifying similarities among users, the system predicts and recommends food items tailored to individual preferences. The proposed solution enhances user satisfaction by aligning services with personal tastes while also streamlining operational workflows. This paper outlines the system architecture, details the implementation of AI methodologies, presents performance metrics, and discusses the system’s impact on hospitality service personalization.
Hotel Management, Database System, Booking, Food Orders, Customer Feedback, Owner Information, Data Integrity, Operational Efficiency, Hospitality Industry, Relational Database
M. Ranjani, Dr. P.R.Tamilselvi
Cardiovascular diseases (CVDs) remain the leading cause of mortality worldwide, emphasizing the urgent need for accurate and early detection methods. This research aims to enhance the predictive performance of heart disease detection by leveraging two powerful machine learning techniques: Extreme Gradient Boosting (XGBoost) and Autoencoders. Using the publicly available Cardiovascular Disease Dataset, we develop and evaluate models based on both techniques, with a focus on key classification metrics—Accuracy, Precision, Recall (Sensitivity), Specificity, and F1 Score. The XGBoost model is trained
directly on the dataset, while the Autoencoder is employed for both anomaly detection and feature extraction. The study compares the performance of these models and explores the potential of a hybrid approach combining Autoencoder-based features with XGBoost classification. Results indicate that both models show promise, with XGBoost achieving high classification accuracy and Autoencoders contributing to enhanced feature representation. This work contributes to the growing field of AIassisted medical diagnostics and offers insights into model selection for heart disease prediction tasks.
Heart Disease Detection, Cardiovascular Disease Dataset, XGBoost, Autoencoders, Machine Learning, Medical Diagnosis, Feature Extraction, Accuracy, Precision, Recall, Specificity, F1 Score
Dr.B.Venkatesan, K.Rajesh
The rapid advancement of Artificial Intelligence (AI) has paved the way for sophisticated Human–Computer Interaction (HCI) systems that leverage multiple input modalities to enhance usability and accessibility. This paper proposes a Multimodal AI System integrating real-time hand gesture recognition, natural voice command processing, and customizable manual shortcuts to provide an adaptive and resilient interface for diverse user environments. Using standard webcams and microphones, the system applies computer vision techniques with Mediapipe and Convolutional Neural Networks for accurate gesture detection, and employs speech recognition algorithms for voice command interpretation. Manual shortcut mapping offers a reliable fallback to ensure uninterrupted control. By fusing these modalities at both feature and decision levels, the system achieves robust performance across variable lighting and acoustic conditions. Extensive evaluations demonstrate the system’s efficacy, highlighting improvements in accessibility for differently-abled users and enhanced interaction fluidity in AR/VR, assistive technologies, and productivity contexts. The proposed approach exemplifies the practical integration of heterogeneous data streams within multimodal AI, reflecting contemporary trends in intelligent interactive systems.
Natural Language Processing, Artificial Intelligence, Human–Computer Interaction, Convolutional Neural Networks, Gesture Recognition Module
E.Jansirani, Dr.N.Kowsalya
Now a day, Security is becoming a main concern to maintain confidentiality and integrity of the data. For this purpose, cryptography techniques are used. Cryptography is used to encrypt the data into non readable format and decrypt it again into readable format when needed using specific key. Also for business perspective, cloud computing is very useful. Cloud Computing has become one of the most talked about technologies in recent times and has got lots of attention from media as well as analysts because of the opportunities it is offering. It has different meaning to different uses according to their need. Cloud computing provides an illusion to the customers of using infinite computing resources that are available from anywhere, any time on demand. Cloud computing provide secure data transmission. Security of data becomes a large concern to insure various attribute like integrity, confidentiality, authentication etc., Asymmetric cryptography techniques like DSA,RSA, Elgammal and ECC plays a major roles in protecting the data in those application which are running in a network environment. In this paper providing comparative analysis on various security algorithm which are already available.
Cryptography, Cloud Computing, Dsa, Rsa, Elgammal, Elliptic curve cryptography, Encryption and Decryption, Asymmetric cryptography