Browse Items (14421 total)
Sort by:
-
Risk-Based Authentication System Using Hierarchical Sub-Feature-Based Model-(HSFBM)
Password-based authentication system recently has been more secure as risk-based authentication system (RBA) is indentured. The RBA system monitors the parameters extracted during the user login process, and based on the proposed model, the system raises a multi-factor authentication to the user. As the vulnerability has increased concerning passwords, fingerprints easy access to any web application may result in a security flow. Several best practices have addressed these issues, but the security threats have been challenging during the initial login sessions. Hence, this paper proposes a novel method for an effective risk identification method during the initial login phase using a hierarchical sub-feature-based model for different categories of users in an RBA system. The FAR is comparatively better in our proposed model, with minimal re-authentication requests for the user. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
A Comprehensive Meta-Analysis on Animal Identification Using Machine Learning and Deep Learning
Artificial Intelligence (AI)-based models have shown promising results in the identification of animals breeds. The surge in the development of new models has opened up new avenues for computer vision. The growing need to achieve cent percent accuracy in the prediction, identification and classification of data/images has motivated researchers to develop innovative strategies seamlessly. The results of various AI models are analyzed in terms of their classification accuracy. EfficientNet-B0 provided an accuracy of 95% in cat breed identification. InceptionV3 deep learning model reached the maximum accuracy of 96.75%, 96.57%, and 100% on dog, goat, and pig breed identification, respectively. ResNet attained an accuracy of 85.77% on snake species identification. This article provides an in-depth analysis of animal classification/species identification models. The inferences drawn out of this literature review would help the researchers in the selection of an ideal AI model to develop an automated animal classification model. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Unveiling Sentiment Trends: An Approach to Utilize Machine Learning in Studying User Activities on New Social Applications
Sentiment analysis is the examination of textual data to determine the writer's attitude, which can be positive, negative, or neutral. In the context of social media analysis, sentiment analysis is peculiar as it helps to identify trends in large amounts of data that are posted by social media users. In the case of sentiment analysis algorithms, the text is categorized into positive, negative, and neutral. Classification of sentiments involves the use of several algorithms such as the decision tree, support vectors, and neural networks. In other words, the paper intends to determine the users sentiment using the decision tree model. Some of the common data sets that have been utilized in this study include the COVID-19 pandemic data, movie reviews, and product ratings. What is tried to be accomplished in this type of case is to determine the efficiency and stability of the decision trees, as well as their optimum success region. Based on the results, it can be pointed out that the accuracy is the highest for the COVID-19 Tweets dataset when referring to the simulation model, which is 98%; hence, the decision tree is best used in the context of the health sector. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Integrating LightGBM and XGBoost for Robust Plant Disease Classification: A Homogenous Stacking Approach
In addressing the critical challenge of early and accurate plant disease diagnosis, this study explores the application of a novel homogeneous multi-layered stacking model utilising Light Gradient Boosting Model (LGBM) and Extreme Gradient Boost (XGB) for the detection of plant diseases. Traditional approaches often rely on basic stacking methods; however, this research seeks to explore the intricacies of altering model architecture, combining the strengths of LGBM and XGB classifiers to build a highly accurate and efficient disease detection system. Comprehensive evaluations were conducted using metrics such as AUCROC curves, Confusion matrix and F1 scores. The ROC curve for the stacked model demonstrated superior performance with a score of 85.12%, compared to 83.09% for the single LightGBM model used for comparative analysis. The future scope of ML in agriculture includes integrating such models with real-time monitoring systems and expanding its applications to diverse crops and environments. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Circulating Current Control in Modular Multilevel Converter Using a Super Twisted Sliding Mode Control
A Modular Multilevel Converter (MMC) has gained widespread popularity due to its notable advantages, including high efficiency, flexibility, minimal harmonic distortion, lower power device ratings, modularity, and scalability. Despite these benefits, the MMC encounters challenges such as voltage balancing and circulating currents. The primary objective is to balance capacitor voltages in each submodule without excessive power electronic device switching. To address this, a controlled switching frequency technique is employed to reduce the average switching frequency of the devices, mitigating voltage imbalances. However, challenges persist, notably the introduction of even-order harmonics with the second harmonic being particularly prominent in inner current. These even order harmonic components contribute to increased power losses, device stress, and potential system instability. Conventional methods of suppression of circulating currents have limitations in harmonic elimination and complex in implementation. This paper proposes a super twisting sliding mode control (SMC) to concurrently manage balancing of capacitor voltages and control of circulating currents in the MMC. The controller effectively reduces the harmonic components, with a focus on the dominant second harmonic in inner currents. Phase disposition (PD) PWM is utilized for MMC phase control, and the proposed controller, initially implemented for a single-phase MMC, demonstrates potential extension to three-phase systems. Comparative analysis with conventional PI controllers and the presentation of results in the MATLAB/Simulink environment validate the efficacy of the proposed super twsiting SMC controller. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Benchmarking Ensemble Methods: Stacking, Hard Voting, and Soft Voting
This study evaluates three ensemble techniquesbasic stacking, hard voting, and soft votingfor predicting diabetes onset using the Pima Indians Diabetes dataset. While traditional methods often focus on single models, this research emphasizes the benefits of combining models like Cat Boost, random forest, logistic regression, linear discriminant analysis, and gradient boosting classifier (LightGBM) within ensemble frameworks. The models were rigorously assessed using metrics for evaluation such as AUC-ROC curves, confusion matrices, F1 scores, etc. The advanced calibrated model achieved the highest performance, with an accuracy of 90.10%, precision of 90.32%, recall of 81.16%, and an F1 score of 85.50%. The soft voting model also delivered strong results, with an accuracy of 89.06%, precision of 87.50%, recall of 81.16%, and F1 score of 84.21%. In comparison, the hard voting model recorded an accuracy of 88.02%, precision of 85.94%, recall of 79.71%, and F1 score of 82.71%. These results highlight the potential of advanced ensemble methods to enhance predictive accuracy. Future work could involve integrating these models with real-time monitoring systems for improved healthcare diagnostics and applying them to diverse datasets and medical conditions. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Benchmarking Ensemble Methods: Stacking, Hard Voting, and Soft Voting
This study evaluates three ensemble techniquesbasic stacking, hard voting, and soft votingfor predicting diabetes onset using the Pima Indians Diabetes dataset. While traditional methods often focus on single models, this research emphasizes the benefits of combining models like Cat Boost, random forest, logistic regression, linear discriminant analysis, and gradient boosting classifier (LightGBM) within ensemble frameworks. The models were rigorously assessed using metrics for evaluation such as AUC-ROC curves, confusion matrices, F1 scores, etc. The advanced calibrated model achieved the highest performance, with an accuracy of 90.10%, precision of 90.32%, recall of 81.16%, and an F1 score of 85.50%. The soft voting model also delivered strong results, with an accuracy of 89.06%, precision of 87.50%, recall of 81.16%, and F1 score of 84.21%. In comparison, the hard voting model recorded an accuracy of 88.02%, precision of 85.94%, recall of 79.71%, and F1 score of 82.71%. These results highlight the potential of advanced ensemble methods to enhance predictive accuracy. Future work could involve integrating these models with real-time monitoring systems for improved healthcare diagnostics and applying them to diverse datasets and medical conditions. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Automation of Breast Cancer Diagnosis and Treatment Using Machine Learning
Breast cancer remains a major global health challenge, with the complexity of managing diverse diagnostic tests often hindering timely and accurate detection. This system proposes a solution by unifying various test results, such as imaging, biopsy and genetic data, into a single platform that leverages machine learning (ML) to predict the likelihood of breast cancer. The platform features an intuitive dashboard that visually represents deviations from normal values, enabling healthcare providers to make informed decisions for early detection and treatment planning. In addition, the system includes an interactive chatbot powered by natural language processing, which assists both doctors and patients by interpreting test results, explaining predictions and offering real-time suggestions for treatment options. This comprehensive approach not only integrates ML models to enhance diagnostic accuracy but also provides real-time updates and alerts for critical changes in patient data. By consolidating fragmented information and incorporating predictive analytics, the system aims to improve the precision of cancer monitoring and offer personalized treatment guidance. About 98% early detection accuracy is achieved to do decision-making processes better which leads to efficient treatment planning. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Deep Reinforcement Learning with Meta-Learning and Signal Bands for Indian Equity Portfolio Management
The portfolio is a collection of assets belonging to an investor. Managing the portfolio depends on the goal of the portfolio management. This paper proposed a new portfolio managing technique using a deep reinforcement learning framework combined with meta-learning and signal bands to optimize the returns and risk of the Nifty 50 index. The objective is to maximize portfolio returns by minimizing the risk, portfolio volatility, and drawdowns with constraints of transaction cost, maximum and minimum allocation, and availability of cash and holdings. The model executes the actions of buy, sell, and hold with the constraints, and the model executes any of those actions depending on the situation and model training. Proposed model recorded a 4.68 Sharpe ratio and 7.53 Sortino ratio while training the model. While testing the model, it recorded a 4.5 Sharpe ratio and 7.64 Sortino ratio, which aligns with the aim to achieve a higher Sortino and Sharpe ratio to build a robust model for risk-adjusted returns. Proposed approach aims to create a strong model for a portfolio management system that adapts to dynamic market conditions and optimizes investment strategies by integrating these techniques. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Evaluating Social Priorities in Environmental Social Governance for the BFSI Sector: A Fuzzy Analytic Hierarchy Process Perspective
As global financial systems evolve, the Banking, Financial Services, and Insurance (BFSI) sector faces increasing pressure to balance financial performance with Environmental, Social, and Governance (ESG) obligations. However, integrating social factors such as employee welfare, community engagement, customer satisfaction, and diversity and inclusion remains challenging due to their subjective and often intangible nature. This study addresses this issue by applying the Fuzzy Analytic Hierarchy Process (Fuzzy AHP) to evaluate and prioritize social factors within the ESG framework. The Fuzzy AHP method, which combines traditional AHP with fuzzy logic to manage uncertainty in expert judgments, was used to gather and analyze input from BFSI sector experts. The study assessed the relative importance of social factors through structured pairwise comparisons, providing a clear hierarchy of priorities for BFSI institutions. The results reveal that employee welfare and customer satisfaction emerged as the most critical social aspects, reflecting stakeholder expectations and regulatory pressures. By focusing on these key areas, BFSI institutions can enhance their ESG performance and meet sustainability goals. These findings offer actionable insights for decision-makers in the BFSI sector, allowing them to better allocate resources to social initiatives that not only satisfy regulatory requirements but also contribute to long-term business value and societal impact. This study underscores the importance of prioritizing social factors in sustainable strategies and provides a robust framework for navigating the complexities of ESG integration. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Predicting Football Players Market Value via Machine Learning
Football, arguably the most popular sport in the world, has become much more than just a sport, it is a multibillion-dollar industry with its center in Europe. Every year millions of euros are spent in transfer window to buy and sell players and a common theme that has been seen is players not living up to the price the clubs paid for them. This research aims to predict football players market values using machine learning techniques. Departing from traditional methods that broadly categorize players into positions like Goalkeeper, Defender, Midfielder, and Forward, this study provides a more nuanced approach by classifying players into specific roles such as Center-back, Full-back, Defensive Midfielder, Attacking Midfielder, and Winger. By incorporating performance metrics tailored to each position and weighing the performance indicators based on the relevance to that specific position, the research aims to provide a robust method to predict players market value within a negotiation tolerance range. Using data from the past three seasons, including detailed player performance statistics and contractual details, models were developed to assist clubs in making data-driven transfer decisions. Machine learning algorithms, including Random Forest Regressor and Light GBM, were utilized, with RMSE and R2 Score as evaluation metrics. Both algorithms demonstrated robust performance, with some positional models predicting market values within an acceptable error range of 312million, enabling clubs to negotiate transfer fees with greater precision based on empirical evidence of player performance. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
AI-Powered Wheels: Machine Learning Approaches for Predicting Used Car Prices
Predictive analytics is now an essential tool for dealers, buyers, and sellers due to the used car markets increasing need for precise pricing models. This study compares the capability of Logistic Regression, Random Forest, Linear Regression, Support Vector Machine (SVM), and Gradient Boosting Machines (GBM) for predicting used car pricing. The results demonstrate that Random Forest and Gradient Boosting scored the best accuracy (87%), with Random Forest also demonstrating better precision (90%). Logistic and Linear Regression both achieved comparable accuracy of 85%, with precision scores of 88% and 89%, respectively. SVM, while significantly less accurate (83%) and precise (86%), produced comparable results for high-dimensional data. In terms of training time, Linear Regression (0.0089 seconds) and Logistic Regression (0.0094 seconds) were the fastest, whereas Gradient Boosting (0.8312 seconds) and Random Forest (0.4766 seconds) took much longer. These results demonstrate a trade-off between model complexity, accuracy, and computing efficiency, with simpler models performing better in terms of speed and ensemble models doing better in terms of prediction accuracy. This study presents practical insights to help stakeholders choose machine learning models for predicting used car prices depending on their specific requirements. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Proposing an AI-Enabled Waste Segregation System for Domestic Settings
This paper proposes an innovative AI-based system for automated domestic waste segregation. Utilizing Teachable Machine and MobileNet, the system accurately categorizes waste into dry and wet components, laying the foundation for sustainable waste management practices. Embedded in a Raspberry Pi 4, the system integrates real-time image processing with various sensors to streamline the sorting process. While the model has been simulated due to budgetary constraints, future implementation envisions real-world application. Potential advancements include expanding the dataset, enabling multi-category waste classification, and exploring low-power alternatives. This research contributes to the evolving landscape of smart waste management, addressing environmental sustainability and the pressing need for automated, efficient waste segregation at the domestic level. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Human Activity Analysis Based on Smartphones and Smart Glasses
The study explores the application of smart glasses and smartphones to study human behavior. Through ensemble and deep learning methodologies, the study seeks to autonomously scrutinize data from each device to improve accuracy and resilience in activity identification. The methodology adopted entails the utilization of distinct models for data derived from smartphones and smart glasses, as opposed to amalgamating attributes, to acquire distinctive insights into user activities. The study outcomes demonstrated promising results, showcasing elevated precision in activity recognition across various machine learning models. Comparative analyses with prior research work reveal enhancements in algorithmic efficacy. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
A Multi-criteria Decision-Making Approach for Prioritising Customer Churn Factors in OTT Video Platforms
Over-the-top (OTT) platforms have revolutionised media consumption by providing on-demand streaming services. Despite their growing popularity, customer churn remains a significant challenge for the platform. This research paper analyses the factors affecting customer churn in OTT video platforms. The factors are identified through unstructured interviews with industry experts and an extensive literature review. This research paper employs a novel approach to prioritising customer churn factors by incorporating multi-criteria decision-making (MCDM) techniques like AHP and fuzzy AHP. The importance of each customer churn factor is measured based on the analytic hierarchy process (AHP) and fuzzy AHP to develop a hierarchy of churn factors. The MCDM analysis results indicated that the content variety and recommendation system, video streaming issues, and high subscription prices are the most significant factors that cause customer churn. Through comprehensive analysis, the study aims to provide insights for OTT service providers to enhance customer retention and mitigate churn rates. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Personalized Medicine Recommendation Through Genomics AI
Every individual is genetically different. This highlights the importance of exploring the possibility of personalized medicine tailored for each persons unique genetic profile. Due to this genetic variability, people have varied responses to the same medication. Thus, it is all the more necessary to withdraw from traditional medication and adopt personalized medicine. Using genomic data analysis techniques, healthcare specialists will be able to distinguish the minute genetic differences to produce unique therapies for each patient. This paper traverses the development of an algorithm to integrate the patients genomic data with his medical history. Personalized treatment can be recommended based on the inferences using genomic AI. These insights derived from the algorithm can be scrutinized by decision support systems. This step ensures the reliability of the prescribed personalized treatment and confidence of patients on the medication. Each time the model predicts medication compositions and dosages using genomic data, it becomes more accurate improving therapeutic results. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Prioritization of Challenges in EdTech Platform to Enhance User Continuance Intention: A Multi-criteria Decision Making Approach
In the rapidly evolving digital education landscape, EdTech platforms face significant challenges that impact user continuance intention. This study employs a fuzzy logic approach within the Multi-criteria Decision Making (MCDM) framework to identify and prioritize these challenges, ensuring the long-term sustainability of EdTech solutions. Key challenges were identified through an extensive literature review and unstructured interviews with eight industry experts. The fuzzy AHP technique was used to rank these challenges, providing a structured approach for EdTech companies to enhance user continuance intention and platform effectiveness. Results reveal Personalization (32.90%) as the most critical factor, followed by Data Privacy and security (20.86%) and User Interface (12.02%). Addressing these prioritized challenges can significantly improve user engagement and contribute to the development of inclusive and accessible educational technologies. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
A Smartphone Based Robust Model for Intelligent Prediction of Thyroid Disease
The thyroid is an integral organ that exerts vital functionality in regulating the bodys functionality in many different ways. Intelligent thyroid disease prediction is very crucial. Through this paper we will describe a smartphone model for predicting whether a patient has thyroid disease based on certain attributes whose values are given by the user. Here, a profound machine learning algorithm named Random Forest will help us predict thyroid disease. By exercising Random Forest within a mobile application, the users can learn about any potential thyroid disease disorder and thus seek medical help in time. This will help improve both the patients and the medical providers quality of care and the help received. With the integration of the Random Forest algorithm along with the mobile application it will prove to be a vital tool which will help everyone in the healthcare fraternity because of its availability and accessibility. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025. -
Enhanced Random Forest-Based Model forFlood Detection andClassification
Flooding is one of the most devastating natural disasters globally, causing extensive damage to infrastructure, the environment, and human lives. With increasing occurrences due to climate change, accurate classification and analysis of flood imagery are essential for early detection, damage assessment, and post-disaster recovery. Reliable flood classification systems are critical for early warning, resource allocation, and mitigation efforts, helping to minimize the impact on affected regions. Remote sensing and computer vision techniques, including the Bag-of-Visual-Words (BOV) model, offer powerful tools for interpreting flood images by categorizing and identifying flooded regions across vast and complex terrains. This paper presents a modification of the standard Random Forest algorithm to enhance the accuracy of image classification within a Bag-of-Visual-Words (BOV) model. The modified Random Forest achieves better adaptability and performance across flood image datasets by introducing flexibility in parameter tuning through custom hyperparameters and automatic grid search. This modification addresses challenges in balancing efficiency and accuracy for classifying high-dimensional image data sets. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026. -
Disease Identification for Tea Leaves Using Explainable Artificial Intelligence
Infection can consequently reduce both quality and yield, and causes major threats to tea production round the world. It is therefore sometime difficult to achieve fast, reliable, and precise identification of disease in tea plants and hence the need to embrace new methods of disease identification. To enable realisation of accurately understandable models for classification of the diseases in tea leaves, Explainable Artificial Intelligence (XAI) approaches are applied in this work. In order to train and test machine learning models, we collected a set of repos of high-resolution images of tea leaves affected by various diseases along with meta information. CNN models were trained with the help of our approach and adopting XAI tools as tools for explanation of predictions. From this study, the field of agricultural AI is benefitted from the illustration of how XAI might enhance disease management strategies in tea agriculture. The results demonstrate an accuracy of 87.85%, with precision, recall and F1-scores ranging between 0.78 and 0.95 across different classes. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.