Browse Items (16481 total)
Sort by:
-
QuiltCraft: A Computer Vision Framework for Sustainable Upcycling Waste Textiles into Artistic Quilts
QuiltCraft represents a pioneering approach where we aim to mitigate the pervasive childrens cloth crisis by ingeniously repurposing discarded cloth into captivating and artistic quilts. By harnessing the power of computer vision techniques, predominantly facilitated by using OpenCV and various Python libraries, our innovative system streamlines the intricate process of identifying cloth sizes, segmenting them into manageable pieces, and orchestrating their arrangement into visually stunning quilt designs. This paper explores the methodology, algorithms, and intricate implementation details underpinning the Quilt Craft framework, supplemented by empirical evidence attesting to its remarkable effectiveness. Through the seamless transformation of childrens clothes into captivating works of art, QuiltCraft not only champions the cause of waste reduction but also catalyzes advancing sustainable practices within the fashion industry. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Prophesying Credit Card Frauds Using Predictive and Deep Transfer Learning: A Comprehensive Experimental Perspective
Credit card fraud has become a major issue in the online financial environment, requiring the implementation of smart and automated tools for real-time detection of frauds. Machine Learning (ML) has been an important asset in this area because of its capability to discover underlying patterns, learn new fraud methods, and offer scalable solutions. This study investigates the usage of different classical machine learning and deep transfer learning based on predictive models for credit card fraud detection with a focus on their comparative performance on six important parameters: time elapsed, accuracy, precision, recall, TNR and F1 score. The investigation makes use of a PCA transformed benchmark dataset with a total of 2,84,807 credit card transactions to train models. In depth experimentation is performed using five classical ML models named Random Forest, Logistic Regression, Linear SVM; Non-Linear SVM; XGBoost and four classical Deep Learned models named MLP, Shallow ANN, ID CNN, and LSTM. To enhance experimental validity, prediction capability of four GNN based CNN models such as Boosting-GNN, Jump-Attentive GNN, GNN and PC-GNN are also tested. Deep learning based neural network models are analysed using seven different activation functions and each model is fit using 10 epochs of batch size 512. Testing results point out that overall best performance in classical ML models is shown by Non-Linear SVM with best recall score depicted by ANN on RBG kernel and GPU. In the ensemble category, Random Forest model exhibits overall best performance with best recall for XGBoost. Precision, accuracy and F1 score of random forest and XG boost are highest. Results have shown that in case of Random Forest the accuracy, precision, recall and F1score are 99.9%, 97.7%, 81.9% and 89.12% respectively whereas for XG boost the values for accuracy, precision and F1 score are 99.96%, 92.63%, 83.81% and 88% respectively. Deep Learned models showed high accuracies, however they were significantly utilized computational resources in respect to elapsed time. The study provides a roadmap to financial institutions for efficient model selection while deciding on implementing automated and trustworthy fraud detection systems and helps shape the dynamic world of intelligent financial security solutions to reduce financial losses. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Enhancing Crude Oil Price Prediction with Neural Network Models
A nations economic, social, and national security are all severely affected by variations in crude oil prices, which is a basic energy source. Research on accurately forecasting price changes for crude oil is always progressing. This research presents a forecasting strategy for crude oil pricing using artificial neural networks. The presented model uses standardization techniques to prepare the historical data for the subsequent processes. It is possible to predict future prices by using a Feed Forward Neural Network (FFNN) with four layers. West Texas Intermediate (WTI) and Brent crude oil prices are utilized on a daily, weekly, and monthly basis to demonstration and confirmation. Directional statistic, accuracy of prediction, the model is evaluated using root mean square error and mean absolute error expressed as percentages. Empirical findings confirm that the suggested approach performs better than any of the previous approaches. Additionally, it is noted that the presented method achieved higher prediction in contrast to other methods. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
A Machine Learning Approach to Crude Oil Price Prediction Using Support Vector Machine (SVM)
Crude oil is one of the most important energy sources, and fluctuations in its international prices affect all aspects of the economy. The price of crude oil is influenced by several variables, and the length of time that each component has an effect differs giving an increase in non-linear oil price features. Although it is a complex task, identifying the most essential factor influencing for precise predicting, crude oil prices are essential. Therefore, this study aims to employ a machine learning model to address the intricate relationships among different factors. Primarily, it gathers data regarding West Texas Intermediate (WTI) and Brent crude oil prices as well as macroeconomic variables. Secondly, the data is normalized to prepare it for further analysis. Finally, a crude oil prediction model is constructed using Support Vector Machine (SVM) to predict future international crude oil prices. The daily, weekly, and monthly prices are used to confirm the models efficacy developed using WTI and Brent oil. The models performance is also evaluated by incorporating various combinations of macroeconomic variables to find the most influential factor. Results from experiments show indicates the benchmark model was much exceeded by the developed model and performed very well in terms of prediction accuracy. The findings reveal that selecting the appropriate variables can greatly enhance prediction accuracy. This model has the potential to provide valuable insights for traders, investors, and energy-related enterprises, offering beneficial guidance for decision-making purposes. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Development of an AI Based Framework for Reverse Supply Chain of Pharmaceutical Products
The pharmaceutical reverse supply chain is an integral part of pharmaceutical industry. Due to the complex nature of the process and strict government regulations, it is important to use different AI technologies to increase the efficiency of the reverse supply chain. This research aims to design an AI driven framework for reverse supply chain of pharmaceutical products which would increase efficiency, speed, automate processes and enhance trust among the stakeholders. The framework consists of five modules namely Collection and Sorting Centre, Return Management, Real-time Inventory Management, Disposal Centre, and Data Analytics. In each module different AI technologies have been embedded to increase the efficiency of the system. The proposed framework offers a holistic approach that not only aligns with stringent pharmaceutical standards but also contributes to a more robust, transparent, and environmentally sustainable reverse supply chain. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Attention-Powered Deep Learning for Employee Analytics: A Multi-Model Approach
In the ever-evolving field of human resources analytics, there is the integration of the latest techniques of machine learning that can strongly enhance decision-making. This paper introduces a revolutionary architecture for multi-model neural networks that integrate disparate networks in analyzing the background, development, performance, and engagement of an employee for all key elements of this employee. Each of the processes with attention fine-tunes the importance of features and therefore largely improves the concentration and interpretability of results. These networks are thus ensured of thorough analysis in the form of in-depth evaluation, which enables classification to be discrete and into clear performance categories. Preparation of raw data was also done with much care; we used the Employee/HR Dataset from Kaggle in order to process this raw data before its use in deep learning application. Our proposed architecture outperformed by accurately classifying the employee performance categories, with result showing a high classification accuracy of 86.49% on the test set. This study, therefore, establishes that customized neural network architectures are applicable in supporting organizations in realizing their data driven culture and in making human resource operations more efficient. 2026, Springer Science and Business Media Deutschland GmbH. All rights reserved. -
A Hybrid Deep Learning Model Using U-Net and Vision Transformer for Artificial Intelligence Powered Cervical Stenosis Diagnosis
This study presents a deep learning-based approach for the classification of cervical stenosis using MRI spine images, integrating multiple phases such as preprocessing, segmentation, feature extraction, and classification. A U-Net-based segmentation model effectively delineates key anatomical structures, including the spinal canal, intervertebral discs (IVDs), and neural foramen, improving feature extraction and classification accuracy. Furthermore, ResNet-50 is employed for feature map generation, leveraging deep hierarchical representations to extract meaningful spatial patterns from MRI slices. For classification, a Vision Transformer (ViT)-based model is utilized, taking advantage of its self-attention mechanism to capture both local and global dependencies within MRI images. Unlike conventional CNN-based models, ViT processes MRI scans as patches, enabling a more context-aware analysis of stenotic regions. The model is trained using an 80%20% train-test split and evaluated using standard performance metrics, achieving an accuracy of 92.60%, precision of 90.16%, recall of 95.43%, and an F1-score of 91.56%. These results indicate that the ViT model outperforms traditional CNN-based classifiers in cervical stenosis detection, ensuring higher sensitivity and specificity in real-world clinical applications. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Enhancing Smart Home Security Through Dynamic Authentication for Voice-Activated Commands
The proliferation of voice-activated IoT devices in smart environments has introduced significant security challenges, particularly around unauthorized access and command spoofing. In response, this paper proposes a unique multi-factor authentication framework that combines dynamically generated code words with smartphone-based verification to ensure secure voice command execution. Unlike conventional fixed-passphrase systems, our method generates unique session-based code words and requires real-time confirmation via a trusted mobile device, offering increased resistance against replay and impersonation attacks. The proposed framework is designed to integrate seamlessly with existing virtual assistants and IoT ecosystems, and its applicability extends beyond smart homes to include connected vehicles and industrial IoT systems, offering a scalable and secure authentication solution. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Optimizing Base Station Placement toMinimize Interference forSatellite Terrestrial Networks (STN)
The rapid advancement of 5G and 6G technologies has spurred the development of Satellite-Terrestrial Networks (STNs), integrating terrestrial infrastructure with Low Earth Orbit (LEO) satellites to enable seamless global connectivity. Efficient spectrum allocation and interference management remain major challenges due to limited resources and the dynamic behavior of satellites. This study addresses these challenges by optimizing base station (BS) deployment to enhance spectral efficiency and reduce interference in STN environments. Delaunay Triangulation (DT) is employed to establish initial spatial separation between BSs, followed by gradient descent (GD) for fine-tuned optimization. Simulation results demonstrate that the optimized scenario substantially reduces interference and improves key performance metrics, including SINR, INR, CI Ratio, and received power, with gains ranging from 30% to 400%. These findings, derived from small-scale simulations, indicate the frameworks potential for enhancing STN performance in dense and interference-prone environments and provide a foundation for future research on interference-resilient STN architectures. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Design ofSIRW-Based Self-diplexing Slot Antenna withLow Cross-Polarization Level forMillimeter-Wave Applications
This article presents a miniaturized self-diplexing slot antenna utilizing a substrate-integrated rectangular waveguide (SIRW) for millimeter-wave applications. The proposed antenna is built using a rectangular cavity, a slot on the top conductor, and two microstrip-feed lines. The rectangular cavity is realized for operation at TE110 mode (26.5GHz), which is perturbed for radiation at two distinct frequency bands by loading a slot on the top conductor. Two microstrip-feed lines are employed for excitation of the slot, which generates two operating bands. For better understanding of radiation phenomena, the electric field distribution, equivalent circuit model, and parametric analysis are provided. To further validate the proposed design, a millimeter-wave self-diplexing antenna operating at 22 and 28GHz is built and full-wave simulated. The proposed antenna occupies a smaller footprint area of 0.396?g2 with an isolation greater than 20 dB. Additionally, the suggested antenna offers low cross-polarization level better than ?30 dB. The realized gain and efficiency of the proposed antenna at lower (upper) frequency bands are 5.5 dBi (5.4 dBi) and 90.6% (94%), respectively. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Compact Polarization Insensitive Triple-Band Bandstop Frequency Selective Surface forTerahertz Applications
In this article, a compact triple-band bandstop frequency selective surface (FSS) is designed for terahertz applications. The proposed FSS is a single-layer structure. The unit cell consists of cross-dipoles featuring three distinct arms. The designed FSS offers three bandstop responses centered at 0.34, 0.39, and 0.43 THz with 10-dB stopband bandwidths of 12, 5, and 14GHz, respectively. The filter operation is elucidated using surface current densities and an equivalent circuit model. The unit cell exhibits a surface area of 0.374?00.374?0, where ?0 is free space wavelength at lower resonance frequency. The proposed FSS exhibits excellent stability in incidence angles up to 40? and 80? for TE and TM polarization, respectively. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Blockchain Node Intelligence Based onDecentralized Framework
Blockchain enhances transparency, transaction speed, and governance reliability for organizations such as manufacturing and supply chain organizations by operating in decentralized environments. Supply chain traceability involves tracking products from their origin to customers, requiring transparency, authenticity, and high efficiency. This paper tries to address the performance gain and challenges in blockchain-based supply chain by making efficient use of on-prem and cloud environments in the blockchain network. As the volume of data being generated in blockchain network continues to grow, data security and performance become increasingly critical. Many existing big data security systems rely on controlled third-party providers, making them vulnerable to various security risks. Blockchain technology offers a promising solution by addressing key challenges such as scalability, immutability, trust, data governance, and transparency, thereby enhancing the protection of personal information. This work focuses on assessment of blockchain processing performance through on-prem, distributed, and decentralized environments and possibility of blockchain nodes to have intelligence, optimized processing capabilities by gaining appropriate infrastructure. We analyze the key challenges of blockchain when execution happens in a standalone system than in scalable ones. We tested a few popular mining processes on cloud platforms and a local system to assess execution speed and discuss a suitable platform to host blockchain. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Forecasting Market Turbulence: A Multi-model Study Using GARCH, Random Forest, and LSTM in the Indian Stock Market
The dynamic and unpredictable nature of the Indian stock market presents significant challenges in forecasting return behavior and managing financial risk. This study explores market turbulence through a comparative analysis of three distinct modeling approaches: the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model, Random Forest, and Long Short-Term Memory (LSTM) networks. By analyzing historical return data from Indian Nifty indices, the research captures both linear dependencies and complex nonlinear patterns associated with market volatility. The results highlight the GARCH models strength in modeling conditional volatility, while the machine learning and deep learning techniquesRandom Forest and LSTMexhibit enhanced predictive power in capturing intricate fluctuations in stock returns. The findings suggest that integrating traditional econometric methods with data-driven approaches offers a more comprehensive and accurate understanding of market dynamics. This multi-model framework is valuable for investors, financial analysts, and policymakers aiming to anticipate and navigate periods of heightened market uncertainty. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Synthetic Image Generation for Crop Disease Classification Using Generative Adversarial Networks
Due to biological diversity and unstructured surroundings, agricultural image analysis strives for optimal model performance to better accomplish visual identification objectives. Large-scale, balanced, and ground-truthed image datasets are very helpful, but they are frequently hard to come by, which restricts the creation of very effective models. The identification of plant diseases has benefited enormously from the continuous advancement of deep learning (DL) techniques, which provide a robust tool with incredibly accurate results. However, the efficiency of deep learning models is dependent on the quantity and caliber of labeled data used for training. Precise classification of crop diseases is important for precision agriculture. These models suffer from limited and imbalance datasets especially for rare diseases. The study suggests a framework using Generative Adversarial Network (GAN) for image generation to enhance the classification of diseases. The study employs conditional GAN trained on a PlantVillage and New plant diseases datasets to generate synthetic images of diseased leaves. The images are evaluated using Structural similarity index (SSIM). Then the augmented images are integrated with the CNN classifier to measure the accuracy of disease prediction using synthetic dataset to validate the efficiency of image generation. The Author(s) 2026. -
Metaheuristic Optimization of Deep Learning Models for Land Cover Classification Using Remote Sensing Data
Deep learning techniques have greatly advanced land-cover classification from remote sensing imagery, but their performance depends critically on choosing optimal hyperparameters. Manually tuning hyperparameters (e.g., learning rate, network depth, dropout rate) is time-consuming and often suboptimal. Metaheuristic algorithms offer an automated approach to this problem. In this work, we compare five metaheuristic optimizersParticle Swarm Optimization (PSO), Genetic Algorithm (GA), Differential Evolution (DE), African Vulture Optimization Algorithm (AVOA), and an Enhanced Dipper Throat Optimization Algorithm (EDTOA)for hyperparameter tuning of convolutional neural networks (CNNs), a ResNet-50, and a U-Net. We evaluate these methods on two benchmark land-cover datasets: EuroSAT (patch-level multispectral image classification) and DeepGlobe (pixel-wise satellite image segmentation). Our data preprocessing includes normalization, data augmentation, and computing spectral indices (e.g., NDVI) to enrich the feature set. Each metaheuristic searches the hyperparameter space to maximize validation accuracy (for EuroSAT) or mean Intersection-over Union (mIoU) (for DeepGlobe). In addition to predictive performance, we analyze the computational cost (wall-clock time, epochs to convergence, GPU usage) of each optimizer to assess the trade-off between efficiency and accuracy. AVOA and EDTOA achieve the best results on both datasets (e.g., up to 98.5% accuracy on EuroSAT and 56% mIoU on DeepGlobe), outperforming the PSO, GA, and DE baselines while offering favorable cost-performance balance. These findings demonstrate that advanced metaheuristics can significantly improve deep model performance in land-cover classification. Our contributions include a comprehensive experimental comparison of five optimizers, a detailed methodology integrating spectral index features, a cost performance analysis, and reference results to guide future research. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
A Machine Learning Approach to Modelling Sales Performance Based on Social Media Analysis
By address the strategic challenges faced while maximising the advertising effectiveness, this paper examines the predictive relationship between different advertising sales platform and the brand sales performance. Four different supervised machine learning models - Random Forest, XGBoost, LightGBM, and a voting classifier ensemble model are applied to categorise the brand sales performance. The models are trained on a multi-platform advertising dataset. This research focuses on both the accuracy and interpretability of the data to make it easier identifying which platform affects the brand sales the most, which is quite different from the existing approaches that concentrates on finding predictive accuracy. This study aims to help marketers and businesses to make better and well informed data-driven decisions for marketing strategies by finding the most effective advertising platform. Each of the models were evaluated using Classification metrics which includes Accuracy, Precision, Recall and F1 score, along with confusion matrix. Future scope of the paper includes merging the models in real time systems and also by expanding it to carefully examine different time periods and customer groups. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
IoT-Enabled Smart Breath Analyzer for Real-Time Monitoring of Ammonia, Alcohol, and VOC Biomarkers for Early Disease Detection
Real-time monitoring of breath biomarkers such as ammonia, alcohol, and volatile organic compounds (VOCs) is important for early disease diagnosis of metabolic and organ-related diseases. Traditional disease diagnosis techniques are invasive, time-consuming, expensive and not suitable for using in remote areas. To overcome these limitations, this paper proposes the design and development of an IoT-based Smart Breath Analyzer for real-time monitoring and analysis of ammonia, alcohol, and volatile organic compounds (VOC) concentration present in exhaled human breath. The system consists of different kinds of gas sensors connected to an ESP32 microcontroller for measuring gas concentration, which is processed and sent to Blynk application via Wi-Fi for visualization and disease prediction. A trained machine learning model is used in the system which classifies biomarker patterns that may be associated with conditions such as kidney disorders, respiratory issues, or alcohol influence, based on literature-derived thresholds. The system is presented as a proof-of-concept screening tool rather than a clinical diagnostic solution. The results are showed on an OLED display and accessed via a mobile app developed using the Blynk IoT platform. This non-invasive, affordable, and scalable solution improves continuous health monitoring and early diagnosis. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Hybrid Machine Learning Models for Crowdfunding Success Prediction
Crowdfunding has become a viable option for founders and entrepreneurs as an alternative source of funding, where individuals can access a large pool of supporters to fill the funding gap. There are a variety of reasons why it is difficult to predict if crowdfunding project will be successful. As such, types of projects, duration of the campaign, target funding goal, and overall supporter activity are constantly changing. This research, therefore, aims to explore, the use of machine learning for predictive models that quantitatively leverage the historical records of projects on kickstarter, in order to find the success probability. In order to analyze the predictive ability for campaign success, was used a combination of machine learning models - Logistic Regression, Random Forest, XGBoost, LightGBM, and Decision Trees - the models that had the greatest precision were Decision Tree (99.91% acc), and LightGBM (99.90% acc) hence why they were selected. In addition, this research demonstrates how feature selection coupled with ensemble learning can significantly increase predictive potential by providing valuable information for campaign builders, platform operators, and investors who are undertaking crowdfunding projects. These findings indicate that predictive modelling can support campaign design, promote investor trust, and enhance credibility for crowdfunding platforms, through uncovering fraud. Additional measures measuring social media interaction or sentiment analysis could be incorporated to provide information for better predictive models. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Comparative Analysis of GANs and Diffusion Models for Hyperspectral Image Classification
Hyperspectral imaging, which is obtained across numerous spectral bands, presents difficulties in classification due to its high dimensionality and intricate nature. This study provides a comparison of Generative Adversarial Networks (GANs) and Diffusion models regarding the classification of the Indian Pines, Pavia University, and Salinas Datasets, utilizing Multi-Layer Perceptron and Random Forest classifiers. The findings indicate the GANs combined with Random Forest outperform Diffusion models, attaining accuracies of 88%, 96% and 95% respectively. This approach may not outperform the top models, such as HTD-2D-3D-PCNN, but is simpler in structure and more computational efficient. Key recommendations would be real-time processing, edge device optimization, and applications customized to agriculture and urban planning. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026. -
Bibliometric Analysis of Gifts in the Era of E-Commerce: A Data Mining Approach
Gifting is a universal phenomenon. It is deeply connected with history, human culture, social interactions, and economic activities. This study aims to look at the work of various researchers on online gifting. The keywords (gift OR gifts) AND (online OR electronic OR e-commerce OR virtual) were used on the Scopus Database. Bibliometric analysis was conducted on 397 relevant publications, which were filtered and selected from the list of 1398 documents. Analysis through Term co-occurrence map, Network visualization map of terms in title/abstract fields, and Topic trends, among others, was done. Four primary clusters were found in the Term co-occurrence map as well as Network Visualization Map Most of the research was from the USA and China. The multi-disciplinary element of gifting is visible in the analysis. Some of the emerging topics were virtual reality, live streaming, social networking, advertising, and online shopping. The impact of gifts in promotions and marketing showed the potential of gifts as a major tool for marketers. The study was limited to only the Scopus database and gives insights into the evolution of online gifting behaviour. The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.