Browse Items (16481 total)
Sort by:
-
Enhanced Artificial Neural Network for Emoji Sentiment Analysis
Emojis enhance textual communication by conveying emotions and providing contextual richness. This study compares the performance of supervised machine learning models such as Naive Bayes, Logistic Regression, Random Forest, K-Nearest Neighbors (KNN), and Artificial Neural Networks (ANNs) for emoji sentiment classification. A major addition in this study is the enhancement of the ANN model using an informed weight initialization technique, which speeds up convergence and reduces training time while maintaining improved performance. The experimental results showed that the Enhanced ANN (EANN) model obtained 94% accuracy, a 2% improvement over the baseline ANN model, while lowering training time from 45 to 18 units (60% decrease), highlighting the importance of initialization strategies in deep learning. The initialization method helped the EANN network avoid overfitting, resulting in increased generalization and accuracy. Proper initialization balanced the gradients during backpropagation, avoiding gradient issues that limit deep networks. Also, the informed weight initialization guaranteed that the EANN began training closer to an optimal solution, lowering the possibility of becoming confined in suboptimal local minima. The findings from this study contribute to advances in sentiment analysis and text mining, particularly in terms of improving the efficiency and accuracy of deep learning approaches. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Hierarchical Demographic Classification Using Multi-output CNN
This work constructs a multi-output Convolutional Neural Network (CNN) for demographic classification, including age estimation, gender determination, race categorization, and geographical subcategories of South Asia, North Asia, Europe, and Africa. Sub-region classification is further carried out in the Indian subcontinent: Karnataka, Kerala, Tamil Nadu, Punjabi and Kashmiri. The proposed CNN is trained using a heterogeneous dataset and its performance is compared with some robust pre-trained models such as VGG16, ResNet50, InceptionV3, and DenseNet121. The proposed custom CNN architecture makes use of convolutional layers for feature extraction, dropout for regularization, and custom loss weights for balanced multi-task learning. Experimental results are showing high accuracy values of 97% for age estimation, 99.55% for gender classification, and 98.4% for region and sub-region classification. The performance of the model gives a sign of its ability to handle demographic diversity while striving to minimize bias compared to its currently existing solutions. This work provides a foundation for future advancements in hierarchical facial recognition and demographic classification with an emphasis on ethical AI and fairness in real-world applications, including overcoming dataset limitations, exploring hybrid models, and incorporating multimodal data to give more inclusivity and precision. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Opposite Color Multiscale Local Binary Pattern Features for the Prediction of Bread Edibility
Bread is one of the profoundly consumed staple bakery foods by many people in the world. Quality is a remarkable concern as it is a consumable food product. It depends on the raw ingredients and baking process involved during the preparation. After the purchase of the bread, the quality and in turn the shelf life period of the bread may likely to get affected by the storage method. Hence, the edibility of the bread needs to be estimated. Most of the studies do estimate this using sensory attribute measurements like strange odor, crust color, taste, aroma, hard texture and mold formation. On contrary, this study newly attempts to examine the edibility effortlessly through images. A new variant of texture based Local Binary Pattern features is proposed for the prediction of edibility through analysis of hard texture and mold formation. As there is no benchmark bread sample dataset available for the study, a new dataset of 18,513 images is created. It is observed from the experimentation that the proposed Opposite Color Multiscale Local Binary Pattern features provide good estimation on majority voting with reduced number of features through feature transformation and selection. The accuracy obtained is 0.8493 which is comparable with other common variants of local binary pattern features. Multiple classifiers are evaluated during experimental analysis and ensemble approach outperforms well. As this is a contemporary problem addressed in the domain based on images of bread being first of its kind, it is likely to open up new challenges to be undertaken. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Advancing Tibetan Text-to-Speech: Challenges and Innovations
This initiative aims to develop a platform for Tibetan Text-to-Speech (TTS) technology, addressing the significant demand for this technology for the Tibetan language. The main objective of this project is to create a system that is capable of converting text into natural and good quality speech. Through the compilation of Tibetan text-audio datasets, the project meets the increasing demand for technology that preserves oral traditions and allows Tibetans to communicate with other people interested in the language. The process includes the gathering of varied Tibetan text and audio samples, such as news articles, followed by processing of data through cleaning processes and statistical analysis. A benchmark dataset is created to enable the testing of models. The lack of certain resources for Tibetan TTS is addressed by the development of pre-trained machine learning models specific to acoustic modeling, using the adapted FastPitch model for waveform synthesis through the HiFi-GAN vocoder. The existing models were?further trained utilizing features particular to Tibetan phonetics and tonalities. The TTS approach is a key strategy for improving digital accessibility for Tibetan speakers and for safeguarding their cultural heritage; it finds applications in media, education, and communication, thus helping to preserve the Tibetan language in the digital era. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Navigating the Complexities of Indoor Air Quality: Critical Analysis of Key Pollutants and Monitoring Sites
Indoor Air Quality (IAQ) plays a crucial role in human health, as individuals spend a significant portion of their time indoors. Many existing studies overlook the compounded effects of common environmental factors like temperature and humidity, as well as the health risks posed by long-term exposure to indoor pollutants. However, this review tries to explore how modern building materials, and indoor activities impact IAQ, with a focus on key pollutants such as ultrafine particulate matter (PM2.5 and smaller), volatile organic compounds ammonia, formaldehyde, chlorine, and radon. The review emphasizes the need for more research on IAQ, especially in developing regions, heritage settlements, and the need for standardized evaluation approaches. Moreover, there is a need of practical and affordable solutions that integrate advanced sensing technologies, real-time data insights, and easy-to-use systems to enable people to aware and take control of their IAQ. This review highlights the importance of more rigorous research, creative strategies, and strong policy support to address these challenges, particularly for communities in developing regions, crowded spaces, and vulnerable populations. The main motive is to enhance understanding of indoor air pollution and promote the development of healthier, more sustainable indoor environments. The objective is to support SDG-3 Good Health and Well Being, SDG-11 Sustainable Cities and Communities, SDG-13 Climate Action. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
NAFLD Detection Using Natural Gradient Boosting: A Probabilistic Ensemble Approach for Improved Accuracy and Calibration
A growing global health concern, non-alcoholic fatty liver disease (NAFLD) must be accurately and promptly detected to avoid serious complications. This study suggests a model based on Natural Gradient Boosting (NGBoost) for accurate clinical feature-based NAFLD prediction. In contrast to traditional gradient boosting algorithms, NGBoost uses natural gradients to estimate the entire conditional probability distribution of outcomes, which enhances uncertainty quantification and calibration. Using a publicly accessible Kaggle dataset, the models performance was compared to KNN, SVM, and Decision Tree classifiers. According to experimental results, NGBoost outperformed conventional classifiers in terms of precision, recall, and F1 score, achieving the highest accuracy of 92.8%. Excellent discriminative ability was indicated by the ROC curve, and strong generalization ability with minimal overfitting was confirmed by the trainingvalidation loss analysis. These findings demonstrate how NGBoost may be used to support clinical decisions, allowing for earlier detection and treatment. Subsequent research endeavors will investigate the validation of the model on more extensive real-world datasets and broaden its relevance to additional liver-related conditions. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Novel Real-Time Posture Monitoring System Using Signal Processing and Computer Vision Techniques
This paper presents a novel real-time posture monitoring system using signal processing and computer vision techniques to provide accurate feedback on body posture. By measuring key angles between the head-shoulder and shoulder-hip regions, the system identifies deviations from ideal posture. A Butterworth low-pass filter is employed to smooth the posture data, significantly reducing noise and misclassification of sudden movements as poor posture. The proposed systems novelty lies in the integration of signal processing to enhance data interpretation, ensuring that momentary shifts are filtered out, resulting in more reliable classification and feedback. The system was tested in real-world scenarios, demonstrating its ability to offer immediate, high-accuracy posture feedback. Unlike conventional systems that rely solely on raw data, our approach uses smoothed, noise-free data to provide a clearer understanding of posture, making it suitable for deployment in workplaces, home offices, and rehabilitation centers. Future work will focus on multi-joint analysis, duration-based feedback mechanisms for sustained posture deviations, and the impact of camera angle on measurement accuracy. Overall, the system provides a cost-effective and efficient solution for continuous posture monitoring, aiming to improve health and ergonomics across various settings. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Privacy-Preserving Federated Learning Protocol for Secure Analytics of IoT Sensor Data Using Homomorphic Encryption
The proliferation of Internet of Things (IoT) devices has led to massive amounts of sensitive data generation, making data security a paramount concern. Existing methods often struggle with protecting heterogeneous IoT data efficiently, particularly during model training and communication. In this work, we propose a federated learning framework integrated with secure encryption mechanisms to safeguard IoT data during model training and aggregation. Each client device trains a local model using its own sensor data, encrypts the model parameters, and sends them to a server. The server aggregates the encrypted models and sends back the global model for decryption by the clients, ensuring data privacy throughout the process. The proposed framework reduces the unauthorized access risks and also the experimental results demonstrate that the model results in an accuracy of 92% during prediction tasks. The system's encryption overhead was minimal, with only a 7.5% increase in computation time compared to unencrypted federated learning methods. Future work will focus on optimizing the encryption techniques for resource-constrained IoT devices and exploring adaptive security mechanisms powered by machine learning to detect emerging threats dynamically. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Intelligent Systems for Safety and Well-Being: Legal and Ethical Challenges in AI-Driven Healthcare Systems in India
Artificial Intelligence (AI), being one of the most utilized technologies, provides a wide range of opportunities and puts forward major challenges as well. The integration of artificial intelligence into the healthcare sector has the potential to improve the healthcare provided to patients. Artificial intelligence holds great promise for health, but also comes with serious challenges, including unethical data collection, cybersecurity threats, and amplifying biases or misinformation, said Dr. Tedros Adhanom Ghebreyesus, WHO Director-General.The government of India has grabbed all the opportunities to accommodate AI in various sectors and become one of the technologically advanced nations. The healthcare sector is one of these sectors. However, there are also ethical issues, such as informed consent, bias, transparency, andpatient rights, that are being constantly raised. Also, there arelegal issues, such as liability and data protection. Implementation, cost, andaccessibility are some of the practical issues that might arise. The Indian government has taken certain initiatives, such as the release of the National Strategy for Artificial Intelligence by NITI Aayog. These guidelines and unregulated principles are informative but not sufficient. There is a need for a legislative imperative that combines ethical principles with normative standards mandated by overarching laws and sector-specific regulations. This approach aims to protect individual rights, ensure transparency, and foster accountability in AI systems. The regulation should be adaptable and flexible to support innovation while minimizing risks and ensuring human rights and values are protected. This study focuses on the ethical and legal challenges that arise while integrating AI into the healthcare sector in India. The study discusses safety, liability, bias, data privacy, and International best practices while using AI in healthcare. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Study on the Selection of Features, Classifiers, and Resampling in Plant Disease Detection from Leaf Images
Computer vision has become an integral part of modern agriculture. One of the key applications of computer vision is the automatic detection and classification of plant disease from digital images of plant leaves. In this study, we evaluate the discriminatory capability of selected texture features in identifying plant diseases from leaf images. Texture features are extracted from resized raw images. Experiments are carried out with public data sets of five different plants. Through extensive experimentation, two classifiersRandom forest and XGBoost are chosen for the evaluation. The class imbalance problem is addressed with a simple resampling. Resampling considerably improves the prediction accuracy. With the raw input images, the best feature as well as classifier depends on the plant type and the quality of the input images. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Robust Statistical Depth Methods for Medical Data: A Focus on Location Estimation and Classification
In robust statistics, data depth functions are extremely powerful and can provide measures of central tendency beyond the ordinary means and medians. These functions provide a sense of depth to points in multivariate space, providing by default a center-outward ranking of observations, which is resistant to outliers and which can be applied to complex and high-dimensional data. Various data depth processes are considered to determine the most optimal location measure with real and simulated data. The performance of Mahalanobis Depth (MD), Half-space Depth (HSD), Zonoid Depth (ZD), Projection Depth (PD), and Spatial Depth (SPD) are compared on some health datasets including the Pima Indians Diabetes Dataset and the Wisconsin Breast Cancer (WBCD) Dataset. The results of these procedures are studied based on calculated depth values and error rates in the discriminant analysis. The findings suggest that the highest depth values are always exhibited by Spatial Depth (SPD), with better robustness and stability without losing accuracy, thus making it the best option. Nevertheless, Mahalanobis Depth (MD) also performs well, which is why it is highly applicable to the robust statistical modelling. Moreover, a new Generalized Mahalanobis Depth (GMD) has been proposed, based on robust location and scatter estimators to eliminate the weaknesses of classical MD. The GMD is more robust to contamination and is valid with singular or ill-conditioned covariance structures, and to high-dimensional data of relevance to real-world data, achieving lower misclassification rates. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Study on Robust Feature Selection Methods Using LASSO, LASSO Variants and Ridge Regression in Sports
Regularization tools like Lasso have made a substantial progresses in regression modelling, particularly to high-dimensional data and multicollinear data. Whereas Ridge regression uses L2 regularization to address the problem of multicollinearity, Lasso uses L1 regularization to conduct regression and feature selection together. The weaknesses of Lasso under correlated predictors have inspired the creation of a number of improved variants. The present paper will do a comparative analysis of simple Lasso, Ridge and Lasso extensions like Elastic Net, Adaptive Lasso, Group Lasso and Relaxed Lasso on real-world sports data, with special focus on a new implementation of the improved Relaxed Lasso that involves three optimization strategies: systematic grid search, extended lambda sequences, and nested cross-validation structures. The comparison has been done in terms of feature selection, resistance to outliers, and prediction accuracy on Ice Hockey, Cricket, and NBA data. Various measures of errors are used to analyze models: Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and Coefficient of Determination (R2 ). The results have shown that the Enhanced Relaxed Lasso performs best regarding improvements in performance especially in cricket data and still serves as a competitive data in different sporting scenarios. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Penalized Maximum Likelihood Estimation for the Log-Logistic Distribution with Complete Data
Penalized maximum likelihood estimation is specified for estimating parameters of a log-logistic distribution for complete-data situations. This approach addresses the issues of Maximum Likelihood Estimation, wherein Maximum Likelihood Estimation is often unstable when sample sizes are small, and fails with heavy-tailed or asymmetric data. By adding a ridge penalty to the log-likelihood, we derive new score equations, which are solved numerically. The performance is measured for a variety of shape and scale parameters and sample sizes, with bias and Mean Squared Error as the two main measures. The simulation experiment results indicate Penalized maximum likelihood estimation consistently achieves lower bias and Mean Square Error with small sample sizes and particularly strong improvements under skewed or heavy-tailed data. With larger sample size, the differences between Maximum Likelihood Estimation and Penalized maximum likelihood estimation decrease, as we would expect. These results suggest that Penalized maximum likelihood estimation is a viable estimation method using the log-logistic distribution, especially with small or limited datasets. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Geospatial Analysis of Groundwater Recharge Zones in Bengaluru
Urban flooding in cities like Bengaluru results from excessive rainfall overwhelming drainage systems, worsened by rapid urbanisation and the expansion of impervious surfaces. This study investigates the causes and consequences of urban flooding in Bengaluru, highlighting the decline in natural drainage and the encroachment of water bodies. Using QGIS, a geographic information system tool, spatial data from sources like NRSCs Bhuvan portal and USGS were analysed to identify flood-prone areas, drainage networks, and land use changes. The analysis revealed critical flooding zones such as Bellandur, Bommanahalli, and Mahadevapura. The study also emphasises the importance of implementing Best Management Practices (BMPs) and Rainwater Harvesting (RWH) strategies. Land Use and Land Cover (LULC) mapping, soil infiltration data, and rainfall patterns were assessed to understand urban hydrology. The findings stress the need for climate-resilient infrastructure, lake rejuvenation, and improved public awareness to mitigate future urban flood risks in Bengaluru. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Evaluating the Impact of Rainwater Harvesting on Urban Runoff Reduction in Bengaluru
The ground surface changes from undisturbed soils with a natural plant covering to disturbed soils as urbanisation and population pressure increase, amplifying the rapid expansion of impervious surfaces in cities. Without adequate stormwater controls, these impervious surfaces alter the hydrologic cycle by lowering infiltration rates and raising runoff peak flows and volumes. Bengaluru city experiences a more temperate climate throughout the year, and the heavy destruction caused by rains is Indias IT hub. Numerous drains get submerged and saturated after only two days of excessive rain, symbolizing the necessity of harvesting techniques in collecting rainwater and reducing runoff volume. Built-up area dynamics were observed from supervised classification and rainfall data collected from the IMD. Annual collectable rainwater was calculated considering the different cumulative rainfall frequencies, i.e. (25/50/75) %. The overall volume of containers was calculated considering different return periods of (5/10/20) years along with (10/20/30) minute rainfall duration. A 45.7% and 25.8% runoff volume reduction (%) was seen in the study area by considering two cases of maximum daily rainfall and maximum average monthly precipitation, respectively. The calculations indicated a significant capacity for collecting rainwater from impervious surfaces is achievable. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Quasi Z-Source Inverter with Simple Boost and Maximum Boost Pulse Width Modulation Techniques for PV Grid Connection
The voltage-fed quasi Z-source inverter (qZSI) is emerged as a promising solution for photovoltaic (PV) applications. This paper proposes a novel high-gain partition input union output dual impedance quasi Z-source inverter (PUDL-qZSI) for PV grid-connected system. This advanced inverter design achieves exceptionally low shoot-through duty ratios and high modulation index, resulting in a superior output current with reduced total harmonic distortion (THD). To modulate three-phase qZSIs and other equivalent topologies, a variety of modulation schemes may be used, some of which involve two extra reference signals to generate shoot-through state. The simulation is carried out on the MATLAB/Simulink environment with PV-based grid-connected PUDL-qZSI to measure the harmonic distortion and power measurement. The proposed inverter is subjected to two different pulse width modulation (PWM) analysis are simulated and compared to validate the proposed system. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Measuring Customer Perception on Promotion of Tourism Destinations Using AR and VR Applications: Model Testing and Validation
The study aims to propose and develop a model to measure the customer perception toward promotional videos created using Augmented Reality (AR) and Virtual Reality (VR) technologies to promote tourism places. Using judgment sampling, 400 tourists were chosen, all of whom had visited various tourist spots in Visakhapatnam and had seen at least two promotional movies highlighting various tourist attractions using AR/VR technology. A properly written questionnaire was produced ahead of time to gather visitors perception for the qualities of augmented and virtual reality advertising attempts. The study revealed that passengers expect full information and appropriate motivation from digital marketing efforts that promote specific tourist locations using Augmented Reality and Virtual Reality. Furthermore, visitors anticipate high-quality visual and audio features in digital advertising materials for tourist destinations, with the goal of improving the entire customer experience and inspiring future visits. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Impact of Digital Wallets Usage on Digital Banking Experience
The aim of this study is to explore the influence the digital wallets on digital banking experience of the bank customers. The study used e-wallet literacy scale constructs as independent variables and digital banking experience as dependent variable in order to identify the impact of digital wallets on digital baking experience of the bank customers. The primary data was collected from 300 bank customers who are using the digital wallet provided by the bank or any other third-party digital wallet. Structure equation modeling was executed to explore the influence of digital wallets on digital banking experience. The results indicate that Purchase Transactions (PT) and Investment Transactions (IT) are having high influence on digital banking experience followed by Fund Transfer Transactions (FTP), Method of Payment (MOP) having moderate impact, and finally, Credit Payment Transactions (CPT) and Bill Payment Transactions (BPT) are having low influence on digital banking experience. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Common Impactful Assignment (CIA): An Innovative Approach to Reduce Student Stress in WIL Programmes
In spite of its advantages, working professionals enrolled in Work-Integrated Learning (WIL) degree or professional certificate programmes frequently express heightened stress and anxiety attributable to the programme's demands. This paper proposes pedagogical enhancements to alleviate stress in working professionals enrolled in WIL programmes while enhancing the quality of learning. The paper proposes three pedagogical approaches collectively known as Common Impactful Assignment(CIA) to enhance the way assignment components of the courses are evaluated. The first concept advocates for a shared common assignment problem statement across multiple courses in a semester. Building on this, the second concept extends the interconnection of courses across different semesters, crafting unified assignment statements that underscore the programme's thematic cohesion. These ideas, tailored for degree programmes, facilitate a broader understanding of the interdependencies between various courses, fostering a comprehensive knowledge base. For more focused and practical-oriented professional certificate programmes, the third concept suggests a project-based common problem statement replacing the entire programme's assignment components. This overarching project, aligned with the programme's central theme, aims to streamline the interconnected nature of courses in such focused programmes. The paper provides sample assignment problem statements for each scenario, outlining their respective benefits and challenges, and discusses appropriate assessment methods. Recognizing the psychological well-being of learners, the paper suggests a methodology for assessing determinants such as stress, anxiety, happiness, and overall well-being. In evaluating WIL students before and after exposure to these new pedagogies, this pre-post assessment method analyses the psychological benefits of innovative teaching approaches. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Unmasking Fake Reviews: A Machine Learning Perspective
In the age of online shopping and digital information, product reviews play a crucial role in influencing consumer decisions. Fake product reviews are widespread and can mislead consumers, affect brand reputation, distort market dynamics and necessitate the development of robust techniques to identify and combat fake reviews, which is the focus of this project. The advent of machine learning has transformed several industries, and its capability to combat the issue of fake product reviews is equally significant. The problem is because of fake previews customers which are in huge confusion for not knowing what to believe. The goal is to develop a predictive model that can differentiate between authentic and fraudulent product reviews. In this project to detect fake product reviews, three preprocessing methods, two feature extraction techniques, and six classifications algorithms are used. In this proposed system, we acquired better accuracy 88.93% using SVM. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.