检索结果-内蒙古大学图书馆

Enhancing nutritional status prediction through attention-based deep learning and explainable AI

Intelligence-Based Medicine 2025年 11卷 100255-100255页

作者： Santoso, Heru Agus Dewi, Nur Setiawati Haw, Su-Cheng Pambudi, Arga Dwi Wulandari, Sari Ayu Faculty of Computer Science Universitas Dian Nuswantoro Semarang 50131 Indonesia Center for Medical Technology Innovation (CEMTI) Faculty of Engineering Universitas Dian Nuswantoro Semarang 50131 Indonesia Nursing Department Faculty of Medicine Universitas Diponegoro Semarang 50275 Indonesia Faculty of Computing and Informatics Multimedia University Cyberjaya 63100 Malaysia Electrical Engineering Department Faculty of Engineering Universitas Dian Nuswantoro Semarang 50131 Indonesia Biomedical Engineering Department Faculty of Engineering Universitas Dian Nuswantoro Semarang 50131 Indonesia

Accurate and interpretable prediction of child malnutrition remains a critical challenge, as existing AI models often lack the transparency needed for clinical adoption. This study introduces a deep learning framework enhanced with Multi-Head Attention (MHA) for nutritional status prediction, offering a novel contribution through the first direct head-to-head comparison of CNN-MHA and LSTM-MHA to evaluate the effectiveness of spatial feature learning versus sequential dependency modeling in structured anthropometric tabular data. Our framework integrates advanced preprocessing techniques, feature selection, and Explainable AI (SHAP), enabling clinically aligned and transparent predictions. Experimental results on a 9605-sample dataset reveal that CNN-MHA achieves superior performance (99.08 % accuracy) compared to LSTM-MHA (98.91 %), confirming that spatial modeling is better suited for this dataset type. SHAP-based feature attribution further validates WHO-standard z-scores as the most influential predictors, enhancing model credibility for clinical application. Additionally, the study introduces an IoT-enabled anthropometric data acquisition system, enhancing real-time monitoring and scalability. This research represents a significant methodological advancement in nutritional status prediction, addressing key gaps in feature prioritization, accuracy, and interpretability. By bridging the gap between high-accuracy AI and clinical transparency, this study advances AI-driven nutritional monitoring and offers a scalable, explainable framework for public health interventions. Future research should explore multi-modal data integration to further enhance generalizability and real-world applicability. © 2025 The Authors

关键词： Deep learning Explainable AI Machine learning in public health Multi-head attention Nutritional status prediction

来源：评论

学校读者我要写书评

暂无评论

MoDeST: A dataset for Multi Domain Scientific Title Generation

引用

Knowledge-Based Systems 2025年 321卷

作者： Bölücü, Necva Bilge, Yunus Can Çetintaş, Dilber Yücel, Zehra Data61 CSIRO SydneyNSW2122 Australia Institute of Informatics Hacettepe University Ankara06800 Turkey University of Malatya Turgut Özal Faculty of Engineering and Natural Sciences Department of Computer Engineering Malatya44210 Turkey Necmettin Erbakan University Seydişehir Vocational School Department of Computer Technologies Konya42370 Turkey

Title generation is a crucial task in scientific writing, serving as the first point of contact for readers and impacting an article's visibility and citation frequency. In this paper, we present a novel multi-domain and multilingual dataset, MoDeST (Multi-Domain Scientific Title Generation), designed to advance research in scientific title generation. Unlike existing datasets, which are often limited to a single language or domain, MoDeST includes both English and Turkish titles across various academic disciplines, such as social sciences, medical science, science and engineering, and more. We explore the challenges of generating concise and informative titles using different sources (keywords, abstract, and full article) and evaluate the performance of large language models (LLMs) in zero-, few-shot and supervised fine-tuning (SFT) settings. We conduct human evaluation and calculate correlations between human judgement and automatic metrics, identifying the most effective evaluation metric for assessing title quality. Experiments show that fine-tuning significantly improves LLM performance for title generation, with LLaMA-3.1 8B, 70B, Aya-expanse 8B, and 32B achieving scores of 40.12, 45.21, 45.31, and 47.22 for Turkish, and 45.10, 49.10, 40.02, and 48.54 for English, respectively. Moreover, we find that abstract is the most effective input source for generating titles. Additionally, we analyse domain-specific challenges and the impact of cross-lingual generation, highlighting the need for tailored models for different domains. Our dataset, with its broad representation, ensures applicability across various academic disciplines, enhancing its utility for multi-domain and multilingual title generation while also benefiting the broader NLP community and related tasks. © 2025 Elsevier B.V.

关键词： Population dynamics

来源：评论

学校读者我要写书评

暂无评论

Adapting Rule Representation With Four-Parameter Beta Distribution for Learning Classifier Systems

引用

IEEE Transactions on Evolutionary Computation 2025年

作者： Shiraishi, Hiroki Hayamizu, Yohei Hashiyama, Tomonori Takadama, Keiki Ishibuchi, Hisao Nakata, Masaya Yokohama National University Faculty of Engineering Yokohama240-8501 Japan State University of New York at Binghamton Department of Computer Science BinghamtonNY13902 United States University of Electro-Communications Department of Informatics Tokyo182-8585 Japan University of Tokyo Information Technology Center Chiba277-0882 Japan Southern University of Science and Technology Department of Computer Science and Engineering Shenzhen518055 China

Rule representations significantly influence the search capabilities and decision boundaries within the search space of Learning Classifier Systems (LCSs). However, it is very difficult to choose an appropriate rule representation for each problem. Additionally, some problems benefit from using different representations for different subspaces within the input space. Thus, an adaptive mechanism is needed to choose an appropriate rule representation for each rule in LCSs. This article introduces a flexible rule representation using a four-parameter beta distribution and integrates it into a fuzzy-style LCS. The four-parameter beta distribution can form various function shapes, and this flexibility enables our LCS to automatically select appropriate representations for different subspaces. Our rule representation can represent crisp/fuzzy decision boundaries in various boundary shapes, such as rectangles and bells, by controlling four parameters, compared to the standard representations such as trapezoidal ones. Leveraging this flexibility, our LCS is designed to adapt the appropriate rule representation for each subspace. Moreover, our LCS has a generalization bias to produce as many crisp rules as possible. Experimental results on real-world classification tasks show that our LCS significantly outperformed LCSs with popular rule representations in test classification accuracy on up to 17 of the 25 datasets tested. © 1997-2012 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

LEARNING SURROGATE EQUATIONS FOR THE ANALYSIS OF AN AGENT-BASED CANCER MODEL

arXiv

引用

arXiv 2025年

作者： Burrage, Kevin Burrage, Pamela Kreikemeyer, Justin N. Uhrmacher, Adelinde M. Weerasinghe, Hasitha N. School of Mathematical Sciences Queensland University of Technology BrisbaneQLD Australia Department of Computer Science University of Oxford United Kingdom Modelling and Simulation Institute for Visual and Analytic Computing Faculty of Informatics and Electrical Engineering University of Rostock Rostock Germany

In this paper, we adapt a two species agent-based cancer model that describes the interaction between cancer cells and healthy cells on a uniform grid to include the interaction with a third species – namely immune cells. We run six different scenarios to explore the competition between cancer and immune cells and the initial concentration of the immune cells on cancer dynamics. We then use coupled equation learning to construct a population-based reaction model for each scenario. We show how they can be unified into a single surrogate population-based reaction model, whose underlying three coupled ordinary differential equations are much easier to analyse than the original agent-based model. As an example, by finding the single steady state of the cancer concentration, we are able to find a linear relationship between this concentration and the initial concentration of the immune cells. This then enables us to estimate suitable values for the competition and initial concentration to reduce the cancer substantially without performing additional complex and expensive simulations from an agent-based stochastic model. The work shows the importance of performing equation learning from agent-based stochastic data for gaining key insights about the behaviour of complex cellular dynamics. © 2025, CC BY.

关键词： Lung cancer

来源：评论

学校读者我要写书评

暂无评论

Loss Functions in Deep Learning: A Comprehensive Review

arXiv

引用

arXiv 2025年

作者： Elharrouss, Omar Mahmood, Yasir Bechqito, Yassine Serhani, Mohamed Adel Badidi, Elarbi Riffi, Jamal Tairi, Hamid Department of Computer Science and Software Engineering College of Information Technology United Arab Emirates University United Arab Emirates Department of Information Systems College of Computing and Informatics University of Sharjah Sharjah United Arab Emirates Department of Informatics Faculty of Sciences Dhar El Mahraz Sidi Mohamed Ben Abdellah University Fez Morocco

Loss functions are at the heart of deep learning, shaping how models learn and perform across diverse tasks. They are used to quantify the difference between predicted outputs and ground truth labels, guiding the optimization process to minimize errors. Selecting the right loss function is critical, as it directly impacts model convergence, generalization, and overall performance across various applications, from computer vision to time series forecasting. This paper presents a comprehensive review of loss functions, covering fundamental metrics like Mean Squared Error and Cross-Entropy to advanced functions such as Adversarial and Diffusion losses. We explore their mathematical foundations, impact on model training, and strategic selection for various applications, including computer vision (Discriminative and generative), tabular data prediction, and time series forecasting. For each of these categories, we discuss the most used loss functions in the recent advancements of deep learning techniques. Also, this review explore the historical evolution, computational efficiency, and ongoing challenges in loss function design, underlining the need for more adaptive and robust solutions. Emphasis is placed on complex scenarios involving multi-modal data, class imbalances, and real-world constraints. Finally, we identify key future directions, advocating for loss functions that enhance interpretability, scalability, and generalization, leading to more effective and resilient deep learning models. © 2025, CC BY.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Real-time driver drowsiness detection using transformer architectures: a novel deep learning approach

引用

Scientific Reports 2025年第1期15卷 1-33页

作者： Hassan, Osama F. Ibrahim, Ahmed F. Gomaa, Ahmed Makhlouf, M.A. Hafiz, B. Information Systems Department Faculty of Computers and Informatics Suez Canal University Ismailia 41522 Egypt Artificial Intelligence Department Faculty of Computer Science and Engineering King Salman International University (KSIU) South Sinai 46511 Egypt Computer Science and Engineering Department Egypt Japan University of Science and Technology (E-JUST) Alexandria 21934 Egypt Computer Science and Engineering Department National Research Institute of Astronomy and Geophysics (NRIAG) Helwan 11731 Egypt

Driver drowsiness is a leading cause of road accidents, resulting in significant societal, economic, and emotional losses. This paper introduces a novel and robust deep learning-based framework for real-time driver drowsiness detection, leveraging state-of-the-art transformer architectures and transfer learning models to achieve unprecedented accuracy and reliability. The proposed methodology addresses key challenges in drowsiness detection by integrating advanced data preprocessing techniques, including image normalization, augmentation, and region-of-interest selection using Haar Cascade classifiers. We employ the MRL Eye Dataset to classify eye states into “Open-Eyes” and “Close-Eyes,” evaluating a range of models, including Vision Transformer (ViT), Swin Transformer, and fine-tuned transfer learning models such as VGG19, DenseNet169, ResNet50V2, InceptionResNetV2, InceptionV3, and MobileNet. The ViT and Swin Transformer models achieved groundbreaking accuracy rates of 99.15% and 99.03%, respectively, outperforming all other models in precision, recall, and F1-score. To ensure the generalization and robustness of the proposed models, we also evaluate their performance on the NTHU-DDD and CEW datasets, which provide diverse real-world scenarios and challenging conditions. This represents a significant advancement over existing methods, demonstrating the effectiveness of transformer-based architectures in capturing complex spatial dependencies and extracting relevant features for drowsiness detection. The proposed system also incorporates a real-time drowsiness scoring mechanism, which triggers alarms when prolonged eye closure is detected, ensuring timely intervention to prevent accidents. A key novelty of this work lies in the integration of Class Activation Mapping (CAM) for enhanced model interpretability, allowing the system to focus on critical eye regions and improve decision-making transparency. The system was rigorously tested under varying lighting condit

关键词：

来源：评论

学校读者我要写书评

暂无评论

Expectation maximization—vector approximate message passing based generalized linear model for channel estimation in intelligent reflecting surface-assisted millimeter multi-user multiple-input multiple-output systems

引用

PeerJ computer Science 2025年 11卷 e2582-e2582页

作者： Ali, Shoukath K. Philip, Sajan P. Khan, Arfat Ahmad Moses, Leeban Cengiz, Korhan Akleylek, Sedat Ivković, Nikola Department of Electronics and Communication Engineering Presidency University Karnataka Bengaluru India Department of Electronics and Communication Engineering Bannari Amman Institute of Technology Tamil Nadu Erode India Department of Computer Science College of Computing Khon Kaen University Khon Kaen Thailand Department of Electrical-Electronics Engineering Istinye University Istanbul Turkey Institute of Computer Science University of Tartu Tartu Estonia Faculty of Organization and Informatics University of Zagreb Pavlinska Varaždin Croatia

Channel estimation poses a main challenge in intelligent reflecting surface (IRS)assisted millimeter wave (mmWave) multi-user multiple-input multiple-output (MIMO) systems due to the substantial number of antennas at the base station (BS) and the passive reflective elements within the IRS lacking sufficient signal processing capabilities. This article addresses this challenge by proposing a channel estimation technique for IRS-assisted mmWave MIMO systems. The problem of channel estimation is normally taken as a compressed sensing (CS) problem, typically addressed through algorithms such as Orthogonal Matching Pursuit (OMP), Generalized Approximate Message Passing (GAMP), and Vector Approximate Message Passing with Expectation-Maximization (EM-VAMP). EM-VAMP demonstrates better performance only when a Gaussian mixture (GM) distribution is chosen as the prior for the sparse channel, especially at high signal-to-noise ratios (SNRs). To address this, the article introduces the application of generalized linear models (GLMs), extensions of standard linear models, providing increased flexibility in modeling data that deviates from Gaussian distribution. Numerical results unveil that the proposed Its EM-VAMP-GLM is much more robust to the existing OMP, GAMP and EM-LAMP algorithms. Copyright 2025 K et al. Distributed under Creative Commons CC-BY 4.0

关键词： Gaussian distribution

来源：评论

学校读者我要写书评

暂无评论

Beyond Vanilla Fine-Tuning: Leveraging Multistage, Multilingual, and Domain-Specific Methods for Low-Resource Machine Translation

arXiv

引用

arXiv 2025年

作者： Thillainathan, Sarubi Yuan, Songchen Lee, En-Shiun Annie Jayasena, Sanath Ranathunga, Surangika Department of Computer Science and Engineering University of Moratuwa Sri Lanka Department of Language Science and Technology Saarland Informatics Campus Saarland University Germany Department of Computer Science University of Toronto Canada Faculty of Science University of Ontario of Technology Canada School of Mathematical and Computational Sciences Massey University New Zealand

Fine-tuning multilingual sequence-to-sequence large language models (msLLMs) has shown promise in developing neural machine translation (NMT) systems for low-resource languages (LRLs). However, conventional single-stage fine-tuning methods struggle in extremely low-resource NMT settings, where training data is very limited. This paper contributes to artificial intelligence by proposing two approaches for adapting msLLMs in these challenging scenarios: (1) continual pre-training (CPT), where the msLLM is further trained with domain-specific monolingual data to compensate for the under-representation of LRLs, and (2) intermediate task transfer learning (ITTL), a method that fine-tunes the msLLM with both in-domain and out-of-domain parallel data to enhance its translation capabilities across various domains and tasks. As an application in engineering, these methods are implemented in NMT systems for Sinhala, Tamil, and English (six language pairs) in domain-specific, extremely low-resource settings (datasets containing fewer than 100,000 samples). Our experiments reveal that these approaches enhance translation performance by an average of +1.47 bilingual evaluation understudy (BLEU) score compared to the standard single-stage fine-tuning baseline across all translation directions. Additionally, a multi-model ensemble further improves performance by an additional BLEU score. © 2025, CC BY.

关键词： Neural machine translation

来源：评论

学校读者我要写书评

暂无评论

Stock Price Forecasting in Finance with Artificial Neural Networks

Stock Price Forecasting in Finance with Artificial Neural Ne...

引用

Information engineering (ACIE), IEEE Asia Conference on

作者： Turgut Daloğlu Garibe Osi Murat Aslan Mehmet Beşkirli Cizre District Directorate of National Education Şırnak University Şırnak Türkiye Department of Computer Engineering Şırnak University Şırnak Türkiye Departmen of Medical Education and Informatics Faculty of Medicine Karamanoğlu Mehmetbey University Karaman Türkiye

ISBN: (数字)9798331528409

ISBN: (纸本)9798331528416

This paper examines the applications of Artificial Neural Networks (ANNs) in the field of finance, particularly in stock price forecasting. Artificial neural networks are mathematical models that mimic biological nervous systems and are used to solve complex problems such as financial data analysis. Stock price forecasting is an important topic for investors and finance professionals. Artificial neural networks have been widely used in this field due to their ability to process large amounts of financial data, learn complex relationships and predict future price movements. This study deals with stock price forecasting with artificial neural networks using historical stock prices, economic indicators and other financial data. Neural networks attempt to predict future price movements by learning the complex relationships between these data. Stock price forecasting with artificial neural networks can offer a more flexible and adaptive approach than traditional statistical methods. However, the success of the model depends on factors such as the quality of the data set used, network architecture and training parameters. This study was conducted to understand the potential of artificial neural networks in finance and to evaluate their effectiveness in stock price forecasting. The purpose of this study is to explore how artificial neural networks can be used in financial data analysis, and in particular their potential in stock price forecasting. The study is designed to understand the advantages of applying ANN in finance, to evaluate the stock price forecasting performance of this technology and to provide a new perspective that can influence future investment strategies.

关键词： Training Accuracy Statistical analysis Finance Training data Predictive models Data models Forecasting Long short term memory Testing

来源：评论

学校读者我要写书评

暂无评论

Improving partial weight bearing compliance with a smart insole: Validity, reliability, and feasibility study

引用

Foot and Ankle Surgery 2025年

作者： Avcı, Ender Ersin Akgün, Gazi Uygur, Mehmet Esat Polat, Mine Gülden Demirbüken, İlkşan Marmara University Health Sciences Faculty Department of Physiotherapy Education and Informatics Istanbul Turkey University of Tennessee at Chattanooga College of Engineering and Computer Science (CECS) Department of Engineering Management & Technology /Mechatronics United States Istanbul Medeniyet University Medicine Faculty Surgical Sciences Orthopaedic and Traumatology Istanbul Turkey

Background: Partial weight bearing (PWB) plays a critical role in early stages of rehabilitation following lower-extremity orthopedic surgery. However, it has been reported patients often have difficulty complying with PWB restrictions, particularly in home setting. The newly designed smart insole system aims to address this issue by remotely monitoring the patient's weight-bearing loads and providing real-time feedback for non-compliant weight-bearing to support PWB compliance. The aim of this study was to assess the validity and reliability of the system in measuring weight-bearing loads and to investigate its feasibility in improving PWB compliance. Methods: The research involved two phases: an experimental design to test validity and reliability in a clinical setting, and a randomized controlled design to test feasibility with remote monitoring at home. Validity and reliability were assessed on a healthy subject using simultaneous load measurements from a smart insole and a force plate under three metronome conditions (30,40, and 60bpm). In the feasibility phase, twelve patients who had undergone microfracture surgery for talar osteochondral lesions were recruited at 4 weeks postoperatively. Participants used Android smartphones and met Ferkel & Scaglione classification criteria I, IIA, and IIB. They were randomized into groups with feedback (FB) or without feedback (NFB) and their compliance with PWB limits was monitored remotely 24-hours for one week using the Insole system. Results: The insole demonstrated high reliability (SEM:1.67 N, ICC_1.1:0.97) and, excellent agreement with the force plate (R²=0.9175, p < 0.001). The FB group achieved higher compliance with prescribed PWB thresholds compared to the NFB group. Conclusions: The newly developed insole system is a valid and reliable rehabilitation tool for accurately measuring weight bearing loads. The system's ability to provide continuous remote patient monitoring and real-time feedbac

关键词： Feedback Gait Mobile Applications Patient Compliance Postoperative Care Rehabilitation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：