Annals of Data Science最新文献

Financial Fraud Detection via Multi-teacher Knowledge Distillation from a Structured Feature Perspective 结构化特征视角下多教师知识精馏的财务欺诈检测

Annals of Data Science Pub Date : 2026-10-07 DOI: 10.1007/s40745-026-00738-9

Juxiang Hu, Aihua Li, Qinyan Wei

{"title":"Financial Fraud Detection via Multi-teacher Knowledge Distillation from a Structured Feature Perspective","authors":"Juxiang Hu, Aihua Li, Qinyan Wei","doi":"10.1007/s40745-026-00738-9","DOIUrl":"10.1007/s40745-026-00738-9","url":null,"abstract":"<div><p>Financial fraud detection based on structured data remains challenging due to high dimensionality, feature sparsity, feature redundancy, and the low prevalence of fraudulent samples. This study proposes MTD-FE-AFS, a framework that integrates multi-teacher knowledge distillation, feature enhancement, and adaptive feature selection for financial fraud detection. Specifically, XGBoost and an improved Soft-Label KNN model are employed as complementary teachers to transfer class-discriminative knowledge to the lightweight student model. The framework further incorporates complementary features derived from a Random Forest model, while adaptive feature selection reduces the influence of redundant variables during training. Together, these components improve the ability of the student model to represent and discriminate between complex fraud patterns. MTD-FE-AFS achieves an F1 score of 88.18% and outperforms the main baseline models across multiple evaluation metrics while maintaining higher inference efficiency. These results demonstrate the effectiveness of MTD-FE-AFS for financial fraud detection based on structured financial data.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 6","pages":"1689 - 1714"},"PeriodicalIF":0.0,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230746","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Generalized Two Parameter Estimator in Inverse Gaussian Regression Model 逆高斯回归模型中的广义二参数估计

Annals of Data Science Pub Date : 2026-07-24 DOI: 10.1007/s40745-025-00664-2

Bavneet Kaur Sidhu, Manoj Kumar Tiwari, Vikas Bist, Manoj Kumar, Faisal Ababneh

{"title":"Generalized Two Parameter Estimator in Inverse Gaussian Regression Model","authors":"Bavneet Kaur Sidhu, Manoj Kumar Tiwari, Vikas Bist, Manoj Kumar, Faisal Ababneh","doi":"10.1007/s40745-025-00664-2","DOIUrl":"10.1007/s40745-025-00664-2","url":null,"abstract":"<div><p>This article introduces generalized two-parameter estimators for modelling explanatory variables in positively skewed data using the inverse Gaussian regression model. The standard approach for estimating unknown regression coefficients relies on the maximum likelihood method. Although the maximum likelihood estimator is a powerful estimation tool, its performance can be severely affected by multicollinearity among explanatory variables. When these variables are correlated, the variances and standard errors inflate, reducing estimator precision. As a result, the maximum likelihood estimator becomes unreliable for estimating regression coefficients. Similarly, in the presence of strong multicollinearity, the Ordinary Least Squares estimator also yields misleading results. In this paper, we propose a new estimator and compare it with existing methods, including the Ordinary Least Squares estimator, Ridge Regression estimator, Liu estimator, and Two-Parameter estimator. The discussion further addresses the estimation of biased parameters and examines the necessary and sufficient conditions associated with the proposed approach. We evaluate the performance of the new estimator using the matrix mean squared error criterion and a simulation study that considers various sample sizes, degrees of multicollinearity, parameter settings, and variance levels. To illustrate practical applicability, we analyze the Theoph dataset, demonstrating that the proposed estimator performs more efficiently than the maximum likelihood estimator and other established techniques. The results show that the generalized two-parameter estimators achieve lower mean squared error than maximum likelihood estimators and deliver superior performance compared to existing methods.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 6","pages":"1581 - 1604"},"PeriodicalIF":0.0,"publicationDate":"2026-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230697","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AML Diagnosis via Transfer Learning: A Modified ResNet50 Approach to Leukocyte Classification 通过迁移学习诊断AML：一种改进的ResNet50白细胞分类方法

Annals of Data Science Pub Date : 2026-07-08 DOI: 10.1007/s40745-026-00685-5

Oladayo Remilekun Falola, Muritala Abiola Asafa, Peter Adebayo Idowu

{"title":"AML Diagnosis via Transfer Learning: A Modified ResNet50 Approach to Leukocyte Classification","authors":"Oladayo Remilekun Falola, Muritala Abiola Asafa, Peter Adebayo Idowu","doi":"10.1007/s40745-026-00685-5","DOIUrl":"10.1007/s40745-026-00685-5","url":null,"abstract":"<div><p>Acute myeloid leukaemia a major cause of morbidity and mortality in any part of the world, with the burden greatly increasing in sub-Saharan Africa. This study developed a Convolutional neural network for the detection and classification of AML. Data for this study were collected from an online data repository that consists of peripheral blood smears selected from 100 patients diagnosed with different subtypes of AML at the Laboratory of Leukaemia Diagnostics at Munich University Hospital, and smears from 100 patients found to exhibit no morphological features of hematological malignancies in the same time frame. The study adopted a ResNet50 model architecture as the base model (a convolutional neural network) and was modified with a custom classification head that reduces the 2048-dimensional feature vector to 1024 before the final 10-class mapping, combined with specific freezing strategies tailored for the Single Cell Morphological Dataset of Leukocytes. The feature selection and model formulation were carried out using a machine-learning toolkit in Python. The Modified ResNet50 model was compared with other state-of-the-art architectures like VGG16 and AlexNet, with our Proposed modified ResNet50 model having the best performance with an accuracy of 91%. The results highlighted the effectiveness of the CNN model and demonstrated its proficiency in identifying key morphological classes. The findings underscore the potential of CNN-based tools to assist clinicians in AML diagnosis, particularly in enhancing speed, accuracy, and consistency in resource-constrained settings.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 3","pages":"627 - 647"},"PeriodicalIF":0.0,"publicationDate":"2026-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148458889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Constrained Probabilistic Self-Organizing Map: Enhanced Approach for Degeneracy Prevention 约束概率自组织映射：退化预防的增强方法

Annals of Data Science Pub Date : 2026-06-25 DOI: 10.1007/s40745-026-00710-7

Zakariya Abibi, Zakariae En-Naimani, Khalid Haddouch

引用次数: 0

AI Challenges and Future in the Global South 全球南方的人工智能挑战与未来

Annals of Data Science Pub Date : 2026-06-08 DOI: 10.1007/s40745-026-00706-3

Bolanle A. Ojokoh, Emile R. Chimusa, Oluwafemi A. Sarumi, Oluwarotimi W. Samuel, K.C Santosh

引用次数: 0

Diabetes Prediction Leveraging Various Feature Selection Techniques and Machine Learning Algorithms 利用各种特征选择技术和机器学习算法的糖尿病预测

Annals of Data Science Pub Date : 2026-06-06 DOI: 10.1007/s40745-026-00692-6

Sayed Zabihullah Musawi, Mohammad Farshi

{"title":"Diabetes Prediction Leveraging Various Feature Selection Techniques and Machine Learning Algorithms","authors":"Sayed Zabihullah Musawi, Mohammad Farshi","doi":"10.1007/s40745-026-00692-6","DOIUrl":"10.1007/s40745-026-00692-6","url":null,"abstract":"<div><p>Diabetes is a chronic metabolic disorder with a growing global burden, making early diagnosis essential for effective management. This study proposes a predictive framework that integrates four machine learning algorithms (MLAs)-gradient boosting (GDB), K-nearest neighbors (KNN), gaussian naïve bayes (GNB), and logistic regression (LGR)–with four feature selection techniques (FSTs), namely genetic algorithm (GA), recursive feature elimination (RFE), Pearson correlation coefficient (PCC), and least absolute shrinkage and selection operator (LASSO). The framework was evaluated on the PIMA Indian Diabetes dataset using standard classification metrics, including accuracy, precision, sensitivity, specificity, F1-score, and the area under curve (AUC). Statistical analysis using the Friedman test confirmed significant performance differences among the feature selection techniques, with PCC achieving the best rank. The optimal results were obtained by combining the GDB classifier with GA and PCC-based feature selection, achieving accuracies of 95.14% and 93.52% using only seven and five features, respectively. These results highlight the effectiveness of the proposed framework in enabling accurate early-stage diabetes diagnosis with a compact feature set suitable for clinical decision support systems.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 4","pages":"1033 - 1055"},"PeriodicalIF":0.0,"publicationDate":"2026-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148613209","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Tuberculosis Disease Classification Using Machine Learning-Based Techniques 基于机器学习技术的结核病分类

Annals of Data Science Pub Date : 2026-05-12 DOI: 10.1007/s40745-026-00680-w

Racheal Shade Akinbo, Gladstone Eli Atrakpo, Oladunni Abosede Daramola

{"title":"Tuberculosis Disease Classification Using Machine Learning-Based Techniques","authors":"Racheal Shade Akinbo, Gladstone Eli Atrakpo, Oladunni Abosede Daramola","doi":"10.1007/s40745-026-00680-w","DOIUrl":"10.1007/s40745-026-00680-w","url":null,"abstract":"<div><p>Tuberculosis remains a major global health concern. Early detection is essential to improving treatment outcomes for tuberculosis. This project explores the classification of tuberculosis from medical images using machine learning models, including Random Forest, Support Vector Machine (SVM), and Logistic Regression (LR). These models were trained using extracted features, and the accuracy, precision, recall, and AUC-ROC metrics were used to assess the models. The receiver operating characteristic curve’s area (AUC) and characteristics were used to evaluate model performance. The results revealed that the Random Forest model excelled, achieving an AUC-ROC score of 99%, accuracy of 96%, specificity of 97%, precision of 97%, sensitivity (recall) of 94%, and an F1 score of 95%. In contrast, the Support Vector Machine attained an AUC-ROC of 100%, with an accuracy of 97%, specificity of 98%, precision of 98%, sensitivity of 96%, and an F1 score of 97%. The Logistic Regression (LR) model achieved an AUC-ROC of 100%, an accuracy of 99%, a specificity of 100%, a precision of 100%, a sensitivity of 98%, and an F1 score of 99%. The evaluation suggests that the Logistic Regression model displays superior predictive abilities and overall performance, showcasing its efficacy in learning from the features present in tuberculosis susceptibility in this project data. The performance of these models has showcased how effective they are in classifying tuberculosis from Chest-Xray which will help in the early detection and diagnosis of tuberculosis diseases for radiologists and the medical field at large.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 3","pages":"561 - 574"},"PeriodicalIF":0.0,"publicationDate":"2026-05-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148459163","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Deep Reinforcement Learning for Personalised Music Learning Pathways 个性化音乐学习路径的深度强化学习

Annals of Data Science Pub Date : 2026-03-27 DOI: 10.1007/s40745-026-00689-1

Ni Li, Zixiang Hong

{"title":"Deep Reinforcement Learning for Personalised Music Learning Pathways","authors":"Ni Li, Zixiang Hong","doi":"10.1007/s40745-026-00689-1","DOIUrl":"10.1007/s40745-026-00689-1","url":null,"abstract":"<div><p>Personalised music education necessitates adaptive instruction that responds to each learner’s evolving skill set, emotional state, and stylistic preferences. Yet, most existing intelligent tutoring systems and AI-driven platforms are limited in their ability to address the sequential, expressive, and affective complexity inherent in music learning. This paper presents a novel Deep Reinforcement Learning (DRL) framework designed to overcome these limitations through real-time multimodal learner modelling, affect-cognitive reward optimisation, and policy adaptation via ensemble methods and curriculum-aware exploration. The proposed approach formalises the instructional process as a Markov Decision Process (MDP), utilising a rich, multidimensional learner state that encompasses musical proficiency, practice behaviour, affective signals (including engagement and frustration), and individual learning style. Notably, the framework intro-duces a dual-reward mechanism that balances musical task performance with emotional engagement, thereby promoting sustained learner motivation. Policy learning is further enhanced by employing an ensemble of Actor–Critic agents, coordinated by a curriculum-based scheduler to facilitate smoother skill progres-sion and improved generalisation across diverse learner profiles. Empirical results from both simulated environments and controlled pilot studies indicate that the proposed method significantly outperforms baseline DRL and conventional adap-tive systems in skill retention, time-to-proficiency, and learner satisfaction. This research contributes to ongoing efforts toward emotionally aware and learner-centred music tutoring systems.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 4","pages":"1013 - 1032"},"PeriodicalIF":0.0,"publicationDate":"2026-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148614533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Modeling Interval-Valued Data in the presence of additional information 在存在附加信息的情况下对区间值数据进行建模

Annals of Data Science Pub Date : 2026-03-25 DOI: 10.1007/s40745-026-00690-8

Abdolnasser Sadeghkhani

{"title":"Modeling Interval-Valued Data in the presence of additional information","authors":"Abdolnasser Sadeghkhani","doi":"10.1007/s40745-026-00690-8","DOIUrl":"10.1007/s40745-026-00690-8","url":null,"abstract":"<div><p>Estimating the midpoints of interval-valued data poses significant challenges, particularly when uncertainty and variability are inherent. This paper advances midpoint estimation by leveraging both Maximum Likelihood (ML) and Bayesian methodologies, incorporating various forms of additional information through parameter constraints. In practical applications, such information often naturally arises, enabling the imposition of constraints such as order, rectangular, spherical, and bounded-order conditions on the parameter space. By integrating these constraints, the estimators can exploit structural relationships among parameters to achieve improved precision and accuracy. Through comprehensive simulation studies, we assess the performance of ML and Bayesian estimators under these constrained settings, demonstrating that Bayesian estimators consistently outperform ML estimators, particularly when employing noninformative priors. The constrained Bayesian approach not only results in lower mean squared errors (MSEs) but also yields more precise credible intervals compared to the confidence intervals obtained via ML, showcasing its robustness and efficiency. Furthermore, a real-world climatological case study validates our theoretical and simulation findings, demonstrating the practical utility of the proposed methods for interval-valued temperature data. This example illustrates how integrating additional information–often readily available–enhances the accuracy and reliability of midpoint estimation in multivariate contexts. Our work underscores the potential of structural constraints in symbolic data analysis, providing actionable insights across diverse scientific domains.</p></div>","PeriodicalId":36280,"journal":{"name":"Annals of Data Science","volume":"13 6","pages":"1663 - 1688"},"PeriodicalIF":0.0,"publicationDate":"2026-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s40745-026-00690-8.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

k-Sample Bayesian Test for the Mean k-样本贝叶斯均值检验

Annals of Data Science Pub Date : 2026-03-20 DOI: 10.1007/s40745-025-00668-y

Luai Al-Labadi, Ayman Alzaatreh

引用次数: 0