检索结果-内蒙古大学图书馆

Spatiotemporal dynamics of cyanobacterial blooms: Integrating machine learning and feature selection techniques with uncrewed aircraft systems and autonomous surface vessel data

引用

JOURNAL OF ENVIRONMENTAL MANAGEMENT 2025年 381卷 124878页

作者： Islam, Mohammed Shakiul Dash, Padmanava Liles, John P. Ahmad, Hafez Nur, Abduselam M. Panda, Rajendra M. Wolfe, Jessica S. Turnage, Gray Hathcock, Lee Chesser, Gary D. Moorhead, Robert J. Mississippi State Univ Dept Geosci Mississippi State MS 39762 USA Mississippi State Univ Geosyst Res Inst Mississippi State MS 39762 USA Mississippi State Univ Dept Ag & Bio Engn Mississippi State MS 39762 USA

Cyanobacterial blooms pose significant threats to aquatic ecosystems and public health due to their ability to release harmful toxins, degrade water quality, disrupt aquatic habitats, and endanger human and animal health through contact or consumption of contaminated water. Monitoring phycocyanin (PC), a pigment unique to cyanobacteria, offers a reliable method for detecting and quantifying these blooms, enabling timely interventions to mitigate their impacts. This study aimed to evaluate ten machine learning algorithms (MLAs) for assessing the spatiotemporal variations of cyanobacterial concentrations over an oyster reef in the Western Mississippi Sound (WMS) using remotely sensed imagery from uncrewed aircraft systems (UAS) and in-situ PC concentrations measured by an autonomous surface vessel (ASV). The study further investigated the influence of river discharge and climatic variables on cyanobacterial concentrations using a time-series of cyanobacteria maps. To derive the most accurate PC retrieval model, a comprehensive set of 85 features was initially generated, including individual spectral bands, band ratios, multiple vegetation indices, and three-band indices. Feature selection was performed using a two-step approach that combined Sequential Backward Floating Selection (SBFS) and Exhaustive Feature Selection (EFS). SBFS was first used to iteratively remove features and optimize model performance, while EFS evaluated all possible combinations of the features identified by SBFS to select the best subset. Among the ten MLAs tested, Extreme Gradient Boosting emerged as the top-performing model, achieving an R2 of 0.835, a root mean square deviation of 0.419 mu g/l, an unbiased mean absolute relative difference of 0.176 mu g/l, and an average percentage difference of 18.072 % in retrieving PC concentration. The novelty of this study lies in its data-driven approach to identifying the most suitable machine learning algorithm and feature subsets for PC retrieval,

关键词： Cyanobacterial blooms Phycocyanin Uncrewed aircraft system Autonomous surface vessel machine learning algorithms Feature selection techniques

来源：评论

学校读者我要写书评

暂无评论

A Hybrid machine learning-Based Framework for Data Injection Attack Detection in Smart Grids Using PCA and Stacked Autoencoders

引用

IEEE ACCESS 2025年 13卷 33783-33798页

作者： Tufail, Shahid Iqbal, Hasan Tariq, Mohd Sarwat, Arif I. Florida Int Univ Dept Elect & Comp Engn Miami FL 33174 USA

Cyberattacks, especially data injection attacks, are becoming more common as smart grids are increasingly interconnected. In addition, accurate and unbiased high-quality data is required for model training. Most of the data we collect from the real world is sparse, incomplete, inconsistent, and skewed. To address these issues, we have proposed a framework to detect such attacks in this study. Using a stacked autoencoder architecture, synthetic instances of minority class data were generated. The generated classes address the imbalances in the data to enhance the generalizability of the model and address diverse attack scenarios. Various machine learning algorithms were evaluated, and the Random Forest (RF) model consistently achieved superior accuracy, ranging from 99.32% to 95.89%. In particular, traditional algorithms such as Logistic Regression (LR) exhibited sensitivity to dimensionality reductions, experiencing a 16.96% accuracy drop when the principal components were reduced from all to 10. In contrast, RF demonstrated resilience, with only a 1.67% mean accuracy drop under similar conditions. Both RF and XGBoost (XGB) emerged as standout models, showcasing high accuracy and robust performance even with dimensionality reduction via principal component analysis (PCA). However, reducing PCA components from 10 to 5 led to performance decreases in all models. The Support Vector machine (SVM) Classifier shows the highest accuracy drop of 14.21%. This study shows the importance of understanding algorithmic behavior and data features and how it can impact the performance of ML models. This analysis will strengthen cybersecurity in smart grids and focusing on the critical need for careful feature selection and tuning, particularly for models sensitive to dimensionality reduction.

关键词： Smart grids Principal component analysis Accuracy Autoencoders Random forests Data models machine learning algorithms Dimensionality reduction Computer security Support vector machines Photovoltaic (PV) systems grid-connected PV systems machine learning algorithms random forest autoencoders multi-layer perceptron (MLP) principal component analysis (PCA)

来源：评论

学校读者我要写书评

暂无评论

Still Making Noise: Improving Deep- learning-Based Side- Channel Analysis

IEEE DESIGN & TEST

引用

IEEE DESIGN & TEST 2025年第1期42卷 20-27页

作者： Kim, Jaehun Picek, Stjepan Heuser, Annelie Bhasin, Shivam Hanjalic, Alan Pandora Oakland CA 94612 USA Radboud Univ Nijmegen Fac Sci NL-6500 GL Nijmegen Netherlands Delft Univ Technol NL-2628 CD Delft Netherlands Univ Rennes CNRS Inria IRISA F-35000 Rennes France Nanyang Technol Univ Temasek Labs Singapore 639798 Singapore Nanyang Technol Univ Natl Integrated Ctr Evaluat NiCE Singapore 639798 Singapore

Editor's notes: Side-channel attacks have been undermining cryptosystems for almost three decades. Advances in machine learning techniques have shown great promise in improving the performance and efficiency of side-channel attacks, even on systems with countermeasures. This article provides a systematic approach to applying ML techniques for side-channel attacks. -Jeyavijayan Rajendran, Texas A&M University, USA

关键词： Training Tuning Overfitting Side-channel attacks machine learning algorithms Measurement Linear programming Entropy Deep learning Systematics

来源：评论

学校读者我要写书评

暂无评论

machine learning insights into the strain-stress behavior of nitinol alloys at low temperatures: Utilizing the orange data mining tool

引用

PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART E-JOURNAL OF PROCESS MECHANICAL ENGINEERING 2025年

作者： Mekky, Abdel-baset H. Qassim Univ Coll Sci Phys Dept Buraydah 51452 Saudi Arabia

This study entails a new technique, based on machine learning algorithms, for predicting the strain-stress behavior of Nitinol alloys. The study utilizes Orange data mining software to determine if algorithms like Linear Regression, Random Forest, k-nearest neighbors, and decision tree are related to the effectiveness. The nitinol-based alloy, which has both shape memory and superelasticity, is used in a wide range of biomedical devices, aerospace constructions, and automated apparatus. Temperature variation is a main factor affecting the Nitinol behavior. It is therefore required to carry out an in-depth analysis of strain-stress patterns. Employing machine learning algorithms along with data mining tools allows the exact prediction of the Nitinol alloy performance under any given condition. The research is purposed to establish the relationship between temperature and mechanical strength of Nitinol by analyzing strain-stress curves that were obtained from tensile tests carried out at different temperatures. Looking at overall results, kNN has the highest accuracy of the models upon comparing MSE and RMSE, and has higher R-2 and, therefore stands as the model of high reliability. Results show that machine learning algorithms successfully forecast the mechanical responses of Nitinol under temperature variations, which assists in determining their formability properties and hence, advancing Nitinol applications in various fields. This research demonstrates machine learning uses in developing materials science and engineering knowledge by showing the ability to predict Nitinol alloy behavior, which includes strain-stress characteristics at low temperatures. Furthermore, the research's significance lies in addressing ethical considerations and practical issues, such as the necessity of implementing an organized approach to achieve environmental benefits.

关键词： Nitinol alloys low environmental temperatures machine learning algorithms strain-stress curves orange data mining tool

来源：评论

学校读者我要写书评

暂无评论

Datasets, machine learning Recognition Performance, and Key Issues in Nursing Care Activity Recognition: A Systematic Review

引用

IEEE ACCESS 2025年 13卷 47570-47586页

作者： Yoshida, Yasuyuki Miwa, Hiroyasu Nakae, Satoshi Ogiso, Satoki Ichikari, Ryosuke Okuma, Takashi Univ Tokyo Human Augmentat Res Ctr Natl Inst Adv Ind Sci & Technol Kashiwa II Campus6-2-3 Kashiwanoha Kashiwa Chiba 2770882 Japan Univ Tokyo Natl Inst Adv Ind Sci & Technol SOMPO AIST RDP Cooperat Res Lab Kashiwa Campus 2 Chiba 2770882 Japan

Countries experiencing rapid aging face severe caregiver shortages, necessitating effective workforce optimization. Nursing care activity recognition (NCAR) offers a promising solution by automatically detecting caregiving tasks over time and visualizing workflows. Based on this information, NCAR can facilitate improvements in caregiving processes. This review systematically assessed 32 studies following PRISMA guidelines, analyzing dataset characteristics and recognition performance of algorithms in NCAR to identify factors influencing recognition accuracy. The review revealed significant challenges, including a lack of publicly available datasets, class imbalance, and decreased recognition accuracy in real-world environments. Additionally, sensor configuration and annotation quality were found to significantly impact recognition performance. The limited availability of field data restricts model generalization, highlighting the need for novel algorithms and efficient data collection methods. Expanding datasets, designing appropriate classes, and improving algorithms are critical steps for advancing NCAR research and practical applications. The findings of this study demonstrate the potential of NCAR in optimizing caregiving workflows and mitigating workforce shortages.

关键词： Accuracy machine learning Medical services Training Human activity recognition Aging Accelerometers machine learning algorithms Databases Annotations older adult care systems class imbalance wearable sensors workflow analysis

来源：评论

学校读者我要写书评

暂无评论

Predicting Wastewater Characteristics Using Artificial Neural Network and machine learning Methods for Enhanced Operation of Oxidation Ditch

引用

APPLIED SCIENCES-BASEL 2025年第3期15卷 1351-1351页

作者： Gulshin, Igor Makisha, Nikolay Moscow State Univ Civil Engn Res & Educ Ctr Water Supply & Wastewater Treatment 26 Yaroslaskoye Highway Moscow 129337 Russia

This study investigates the operational efficiency of the lab-scale oxidation ditch (OD) functioning in simultaneous nitrification and denitrification modes, focusing on forecasting biochemical oxygen demand (BOD5) concentrations over a five-day horizon. This forecasting capability aims to optimize the operational regime of aeration tanks by adjusting the specific load on organic pollutants through active sludge dosage modulation. A comprehensive statistical analysis was conducted to identify trends and seasonality alongside significant correlations between the forecasted values and various time lags. A total of 20 time lags and the "month" feature were selected as significant predictors. These models employed include Multi-head Attention Gated Recurrent Unit (MAGRU), long short-term memory (LSTM), Autoregressive Integrated Moving Average-Long Short-Term Memory (ARIMA-LSTM), and Prophet and gradient boosting models: CatBoost and XGBoost. Evaluation metrics (Mean Squared Error (MSE), Mean Absolute Error (MAE), Symmetric Mean Absolute Percentage Error (SMAPE), and Coefficient of Determination (R2)) indicated similar performance across models, with ARIMA-LSTM yielding the best results. This architecture effectively captures short-term trends associated with the variability of incoming wastewater. The SMAPE score of 1.052% on test data demonstrates the model's accuracy and highlights the potential of integrating artificial neural networks (ANN) and machine learning (ML) with mechanistic models for optimizing wastewater treatment processes. However, residual analysis revealed systematic overestimation, necessitating further exploration of significant predictors across various datasets to enhance forecasting quality.

关键词： machine learning algorithms wastewater treatment effluent quality soft sensors

来源：评论

学校读者我要写书评

暂无评论

SENTINEL: Insider Threat Detection Based on Multi-Timescale User Behavior Interaction Graph learning

引用

IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING 2025年第2期12卷 774-790页

作者： Xiao, Fengrui Chen, Shuangwu Chen, Siyang Ma, Yuanyi He, Huasen Yang, Jian Univ Sci & Technol China Dept Automat Hefei 230026 Peoples R China

Insider threats have become a prominent driver behind a myriad of cybersecurity incidents in recent years. Since the threats take place within intranet, traditional security devices located at the network perimeter can hardly detect them. The trust management methods employed within the organization are likewise incapable of intercepting access actions already authenticated with valid credentials. In this paper, we propose a novel insider threat detection method named SENTINEL, which identifies abnormal behavior of insiders and provides fine-grained threat intelligence. We devise a dynamic user behavior interaction graph (BIG), which jointly considers the spatial distribution of user behavioral trajectories among the network topology and the temporal variations of user behavioral profiles. By incorporating a spatio-temporal graph neural network, SENTINEL is able to learn the operation regularities of users at specific times and respective positions in BIG. In order to perceive both the abrupt and persistent threats simultaneously, we conceive a multi-timescale fusion mechanism for detecting users' activities at different timescales. SENTINEL implements a log-entry-level detection without requiring any attack samples during model training. The experiments conducted on widely-used public datasets demonstrate that SENTINEL achieves superior performance while maintaining a relatively low computational overhead compared to the state-of-the-art methods.

关键词： Threat assessment Feature extraction Graph neural networks Security Organizations Image edge detection Network topology machine learning algorithms Hidden Markov models Anomaly detection Insider threat detection graph neural network anomaly detection behavior interaction graph (BIG)

来源：评论

学校读者我要写书评

暂无评论

A novel multiple analysis method for assessing seismic deformations of GRS walls: combinations of FEM, machine learning and performance-based fragility curves

引用

STRUCTURE AND INFRASTRUCTURE ENGINEERING 2025年

作者： Rafsanjani, Ali Akbar Heshmati Razeghi, Hamid Reza Alavi, Seyed Meisam Iran Univ Sci & Technol Sch Civil Engn Tehran Iran

Assessment of seismic deformations of geosynthetic reinforced soil (GRS) walls in literature has dealt with unsolved challenges, encompassing time-consuming analyses, lack of probabilistic-based analyses, ignored inherent uncertainties of seismic loadings and limited investigated scenarios of these structures, especially for tall walls. Hence, a novel multiple analysis method has been proposed, founded on over 257,400 machine learning simulations (trained with 1582 finite element method analyses) and numerous performance-based fragility curves, to promptly evaluate the seismic vulnerability. The conducted probabilistic parametric study revealed that simultaneously considering several intensity measures for fragility curves is inevitable, preventing engineering judgement bias (up to 52% discrepancies in damage possibilities). Up to 75% contrasts between failure possibilities of 8 and 20 m walls, especially under earthquakes with common intensities (e.g. PGA <= 0.3g), raised serious concerns in the application of height-independent designing methods of GRS walls (e.g. AASHTO Simplified Method). Decreases in deformation possibilities were nearly the same due to increasing reinforcement stiffness (J) (1000 to 2000 kN/m) and reinforcement length to wall height ratio (L/H) (0.8 to 1.5);a decisive superiority of J variations over increasing L/H, as a remedial plan. The proposed methodology privileges engineers to swiftly assess the seismic deformations of multiple GRS walls at the design stage.

关键词： Finite element method geosynthetic reinforced soil walls machine learning algorithms performance-based fragility curves probabilistic parametric study seismic analyses

来源：评论

学校读者我要写书评

暂无评论

FairGuard: A Fairness Attack and Defense Framework in Federated learning

引用

IEEE TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING 2025年第3期22卷 2519-2532页

作者： Sheng, Xinyi Yang, Zhengjie Bao, Wei Univ Sydney Fac Engn Sch Comp Sci Sydney NSW 2006 Australia Hong Kong Univ Sci & Technol Hong Kong Generat AI Res & Dev Ctr Hong Kong Peoples R China

Algorithmic fairness, emphasizing that machine learning algorithms should not discriminate against specific demographic groups, has recently gained increasing attention in the context of federated learning. Existing algorithmic fairness algorithms in federated learning primarily focus on mitigating inherent biases within the training data or biases introduced by non-IID data distributions. However, these methods heavily rely on the assumption of trust and cooperation among participants in federated learning, leaving a notable gap in exploring the robustness of algorithmic fairness under adversarial environments. Therefore, our study advocates the exploration of the robustness of algorithmic fairness in federated learning. We first introduce four types of fairness attacks targeting the fairness of federated learning models. Then, we propose FairGuard, an innovative defense framework designed to counter these attacks. FairGuard employs a novel dual-inference algorithm that effectively identifies malicious clients during the global aggregation phase of federated learning. Our comprehensive experiments demonstrate that the introduced fairness attacks significantly compromise the global model's fairness and cannot be effectively mitigated by existing defense mechanisms, while FairGuard exhibits remarkable effectiveness in defending against these fairness attacks.

关键词： Federated learning machine learning algorithms Servers Training data Robustness Training Data models Measurement Collaboration Threat modeling Adversarial attacks and defenses algorithmic fairness in machine learning resilient federated learning

来源：评论

学校读者我要写书评

暂无评论

Analysis of machine learning approaches for predictive modeling in heart disease detection systems

引用

BIOMEDICAL SIGNAL PROCESSING AND CONTROL 2025年 106卷

作者： Mandal, Akshaya Kumar Dehuri, Satchidananda Sarma, Pankaj Kumar Deva Assam Univ Cent Univ India Dept Comp Sci Silchar 788011 Assam India Fakir Mohan Univ Dept Comp Sci Balasore 756019 Odisha India

Heart disease is a leading global cause of death, making early prediction crucial for saving lives. Although the exact causes of heart disease are not fully known, its strong links to high mortality, severe morbidity, and disability highlight the need for advanced predictive technologies. Artificial Intelligence (AI) plays a crucial role in developing these technologies, with machine learning (ML) being a key tool for analyzing data from AI devices. However, creating effective ML-based approaches for predicting heart disease is highly challenging. This study aims to conduct an empirical analysis of twelve (12) promising machine learning approaches with their detailed mathematical analysis for predicting heart disease with optimal accuracy for diagnostic purposes. It details the application development stages, including dataset collection and dataset attribute evaluation. In data analysis, metrics such as accuracy, ROC curves, confusion matrices, and feature importance provide graphical significance, while summary plots of shapely values highlight the significance of features in the applied models, which are used to evaluate the approximate accuracy in detecting heart disease. In conclusion, among the twelve machine learning models evaluated, support vector machines (SVM) achieved the highest performance, with an accuracy of 89%, a receiver operating characteristic (ROC) score of 92%, and a precision-recall curve (PRC) score of 93%.

关键词： machine learning algorithms Heart disease detection system Computational method Feature selection Disease prediction

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：