检索结果-内蒙古大学图书馆

Maximal Information Coefficient-Based Undersampling Method for Highly-Imbalanced Learning

IEEE ACCESS 2025年 13卷 4126-4135页

作者： Qin, Haiou Nanchang Inst Technol Sch Informat Engn Nanchang 330099 Peoples R China Jiangxi Prov Key Lab Smart Water Conservancy Nanchang 330099 Peoples R China

Learning from highly-imbalanced datasets is still a big challenge in the field of machine learning because models created by general learning algorithms are weak in recognizing the samples from the minority class correctly. Undersampling is an alternative kind of methods to deal with imbalanced learning. In this paper, we propose a new undersampling method based on maximal information coefficient (including two algorithms MICU-1 and MICU-2) to rebalance the datasets. In order to evaluate the effectiveness of the method, 20 highly- imbalanced datasets are used for the benchmarks. Results show that compared with other undersampling methods, maximal information coefficient-based undersampling method are competitive in terms of G-mean and F-measure.

关键词： Microwave integrated circuits Generative adversarial networks Noise measurement Machine learning algorithms classification algorithms Training Software packages Shape Sensitivity Sampling methods Imbalanced classification imbalanced learning maximal information coefficient maximal information coefficient-based undersampling undersampling

来源：评论

学校读者我要写书评

暂无评论

Predicting bearing damage in electric motors: a comparative study using combined neural networks

引用

NONDESTRUCTIVE TESTING AND EVALUATION 2025年

作者： Althobiani, Faisal King Abdulaziz Univ Fac Maritime Marine Engn Dept Jeddah Saudi Arabia

The bearing damage within the motors can lead to unexpected failures. This study aims to develop an efficient model for automated prediction of bearing damages within motors, leveraging advanced deep learning techniques combined with traditional machine learning algorithms. The proposed method operates by extracting intricate features from input bearing damage data using a Fine-tuned Convolutional Neural Network (CNN) combined with a Recurrent Neural Network (RNN) model. These extracted features are then integrated into Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), and CART models for predictive analysis. The proposed integrated model was validated using the Paderbone University Bearing Fault Benchmark Dataset, which contains six classes of bearing conditions. The developed Fine-tuned CNN-RNN model achieved an exceptional performance, attaining 96% accuracy across all six bearing damage classes. Comparative analysis with prominent algorithms, including CART (99% accuracy), SVM (96% accuracy), KNN (96% accuracy), and RF (99% accuracy), underscored the superiority of the proposed model's integrated approach. The results demonstrate the efficacy of the Fine-tuned CNN-RNN architecture in accurately predicting and diagnosing bearing damages within motors. This approach has the potential to enhance predictive maintenance and reliability engineering practices, contributing to the improved operational reliability and performance of electric motor systems.

关键词： classification algorithms deep learning fault diagnosis hybrid learning machine learning stability analysis

来源：评论

学校读者我要写书评

暂无评论

An Improved Triangulation Oversampling Method for Processing Unbalanced Data

引用

IEEE ACCESS 2025年 13卷 33655-33664页

作者： Liu, Jingjing Liu, Yefeng Ma, Yanwei Zhang, Qichun Shenyang Inst Technol Liaoning Key Lab Informat Phys Fus & Intelligent M Fushun 113122 Peoples R China Shenyang Inst Technol Dept Basic Courses Fushun 113122 Peoples R China Shenyang Ligong Univ Sch Automat & Elect Engn Shenyang 110180 Peoples R China Shenyang Inst Technol Sch Mech Engn & Automat Fushun 113122 Peoples R China Buckinghamshire New Univ Sch Creat & Digital Ind High Wycombe HP11 2JZ England

In classification tasks, the algorithms heavily depend on big data, yet data category imbalance hinders the model's ability to adequately learn from scarce class samples, impacting its overall learning capacity. This paper addresses the challenge of data imbalance in data-driven classification problems, which introduce an Improved Finite Elements-Synthetic Minority Oversampling Technique (IFE-SMOTE) for data balancing. This method triangulates the sample space leveraging FE-SMOTE, designates three strategic points within each split element based on defined rules, and synthetically generates samples within the linear vicinity of these points. The mathematical expectation of generated samples aligns with the triangle center of the original minority samples, while their variance closely mirrors that of the triangle center, ensuring statistical consistency with the original data. The corresponding theorem is provided and its validity is proved. Numerical experiments confirm the effectiveness of the proposed IFE-SMOTE method. The IFE-SMOTE algorithm classified eight datasets and compared them with other oversampling algorithms using G-mean, F-measure, and AUC. IFE-SMOTE's average scores were 0.9273, 0.9754, and 0.9309, respectively, outperforming FE-SMOTE by 0.0268, 0.0059, and 0.0253, and other algorithms' averages by 0.0493, 0.0169, and 0.0402.

关键词： classification algorithms Optimization Finite element analysis Data integrity Training Technological innovation Redundancy Pressing Overfitting Object recognition Oversampling SMOTE triangulation unbalanced data

来源：评论

学校读者我要写书评

暂无评论

Statistical Feature Extraction and System Identification algorithms for Partial Discharge Signal classification Using Laguerre Polynomial Expansion

引用

IEEE TRANSACTIONS ON DIELECTRICS AND ELECTRICAL INSULATION 2020年第6期27卷 1924-1932页

作者： Janani, Hamed Jayasinghe, Pramoda Jozani, Mohammad Jafari Kordi, Behzad Powertech Labs Inc Surrey BC V3W 7R7 Canada Univ Manitoba Dept Stat Winnipeg MB R3T 2N2 Canada Univ Manitoba Dept Elect & Comp Engn Winnipeg MB R3T 5V6 Canada

In this paper, a novel algorithm for partial discharge (PD) pulse waveform analysis and separation based on orthogonal series expansion using Laguerre polynomials is proposed. A system identification technique is also developed to help increase the accuracy rate of the signal classification. Laguerre polynomial series expansions are applied on the PD pulses to effectively extract important information and well-discriminative features. Laguerre expansion coefficients inherently contain information about the amplitude (intensity) and the relaxation time of a dynamic system and this helps direct characterization of PD pulses. The proposed technique is applied to PD pulses acquired from a laboratory set of multiple PD sources. It is shown that this statistical method can classify partial discharge signals with high accuracies, even when the signals are visually indistinguishable. In this research, two methods are developed for system identification. One is based on a deterministic approach in the form of a recursive formula and the other one is a stochastic method based on a group Lasso methodology. The aim of system identification is to improve the classification accuracy by removing the effect of the measurement system from the observed PD waveforms. The accuracy rate of the proposed algorithm is evaluated using the signals that are captured from simultaneous artificial defects in high pressure SF6 and in low pressure air.

关键词： Partial discharges Pattern classification Feature extraction Dielectric measurement System identification classification algorithms Signal analysis partial discharges feature extraction orthogonal functions signal classification

来源：评论

学校读者我要写书评

暂无评论

Integrated Object Detection and Communication for Synthetic Aperture Radar Images

引用

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2025年 18卷 294-307页

作者： Xu, Zhiping Xu, Deyin Lin, Lixiong Song, Linqi Song, Dan Sun, Yanglong Chen, Qiwang Jimei Univ Sch Ocean Informat Engn Xiamen 361021 Peoples R China Xiamen Univ Key Lab Underwater Acoust Commun & Marine Informat Minist Educ Xiamen 361005 Peoples R China City Univ Hong Kong Hong Kong Peoples R China City Univ Hong Kong Shenzhen Res Inst Shenzhen Peoples R China Jimei Univ Nav Coll Xiamen 361021 Peoples R China Huaqiao Univ Coll Informat Sci & Engn Xiamen Key Lab Mobile Multimedia Commun Xiamen 361021 Peoples R China

In this article, an integrated object detection and communication (IODC) system based on single-stage detection framework is first proposed for synthetic aperture radar (SAR) images in underwater environment. Specifically, the combination of the multiscale feature extraction module and semantic information enhance fusion module for SAR images in underwater environment is designed as the semantic encoder and the results prediction module with anchor-free detection method is designed as the semantic decoder. Considering the multiscale feature, the channel encoder composed by a multiscale fusion module and a redundancy module is designed, and the channel decoder is the inverse of the channel encoder. To adapt to the time-varying and complex wireless environment, an adaptive transmission (AT) mechanism based on attention mechanism and knowledge base is proposed for the IODC system. Moreover, considering the actual application requirements, a lightweight design for the IODC system with the AT mechanism is also conducted. The experiment results on the Sonar Common Target Detection dataset show that the proposed IODC system with AT mechanism can achieve nearly 91% average precision at 25 dB, which means the proposed system can achieve the effective integration of the objection detection and communication for SAR images in underwater environment. In the lightweight design, the model parameters of the proposed system can be reduced by up to 20%, with only a 2.22% sacrifice in performance.

关键词： Radar polarimetry Object detection Transformers Synthetic aperture radar Feature extraction Decoding Wireless sensor networks Remote sensing Deep learning classification algorithms semantic communication synthetic aperture radar underwater environment

来源：评论

学校读者我要写书评

暂无评论

Detection of Small Targets on the Sea Surface Based on High-Dimensional ConvH Classifier

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Wu, Xijie Liu, Tianpeng Liu, Yongxiang Liu, Li Natl Univ Def Technol Coll Elect Sci & Technol Changsha 410073 Peoples R China

This article proposes a viable and easy-to-operate one-class classifier (OCC) of high-dimensional ConvH (high-D ConvH), under the background of feature detection for small targets on the sea surface. Our classifier breaks the bottleneck of current ConvH classifier, i.e., being confined only to the detection within 3-D feature space, by addressing the following two issues: 1) the current ConvH decision criterion is limited to 3-D and 2) the unacceptable learning time of ConvH under high-D space. For the first issue, derived from the half-space description of ConvH, we propose a fast ConvH decision criterion, which can be extended to any dimensional space and owns lower algorithmic complexity. Accordingly, a corresponding fast ConvH decision algorithm is designed. For the second issue, we first give an approach to accelerate ConvH learning, based on which we propose two fast ConvH learning algorithms of different mechanisms. The rigorous mathematical derivations prove that the proposed algorithms are correct and feasible. Combining the above algorithms, the high-D ConvH classifier is obtained, and then a novel feature detection method for small target on the sea surface is presented by integrating the high-D ConvH classifier into existing feature detection process. The experiments on the CSIR datasets reveal that: 1) the high-D ConvH classifier is successfully applied to the 7-D feature space, and it owns shorter learning and decision time than existing methods and 2) the proposed detection method owns superior detection performance compared with its competitors in high-D space.

关键词： Feature extraction classification algorithms Clutter Vectors Feature detection Three-dimensional displays Sea surface Training Signal processing algorithms Convex hulls Convex hull (ConvH) CSIR dataset feature detection high-dimensional sea clutter

来源：评论

学校读者我要写书评

暂无评论

Two-Tiered Privacy Preserving Framework for Software-Defined Networking Driven Defence Mechanism for Consumer Platforms

引用

IEEE ACCESS 2025年 13卷 26684-26694页

作者： Alotaibi, Sultan Refa Alfraihi, Hessa Alruwais, Nuha Maray, Mohammed Ben Miled, Achraf Al-Sharafi, Ali M. Alotaibi, Moneerah Alajmani, Samah Hazzaa Shaqra Univ Coll Sci & Humanities Dawadmi Dept Comp Sci Shaqra 11911 Saudi Arabia Princess Nourah bint Abdulrahman Univ Coll Comp & Informat Sci Dept Informat Syst POB 84428 Riyadh 11671 Saudi Arabia King Saud Univ Coll Appl Studies & Community Serv Dept Comp Sci & Engn Riyadh 11495 Saudi Arabia King Khalid Univ Coll Comp Sci Dept Informat Syst Abha 61421 Saudi Arabia Northern Border Univ Coll Sci Dept Comp Sci Ar Ar 73213 Saudi Arabia Univ Bisha Coll Comp & Informat Technol Dept Comp Sci & Artificial Intelligence Bisha 67714 Saudi Arabia Taif Univ Coll Comp & Informat Technol Dept Informat Technol Taif 21944 Saudi Arabia

Software-defined networking (SDN) is a novel network theory that divides the controller from the network devices such as switches and routers. The integrated SDN structure enables the global network organization and tackles the necessity of present data centers. There are great advantages presented by the architecture of SDN, the hazard of novel assaults is a vital issue and can avert the widespread acceptance of SDNs. The controller of SDN is an essential part, and it is a tempting aim for the invaders. If the attacker effectively acquires the SDN controller, it can transmit the traffic depending upon its desires, causing severe loss to the complete system. Mobile users can access crucial actual services over wireless models such as software-defined networks (SDNs) topologies and the Internet of Things (IoTs). Thus, managing power consumption and system and device congestion turns into a main problem for SDN-based IoT applications. Network intrusion detection systems (NIDSs) are significant devices for identifying and protecting the network landscape from anomalous attacks and malicious activity. Currently, deep Learning (DL) has revealed desired outcomes in a diversity of problems like speech, image, text applications, etc. Whereas numerous works used DL for NIDSs, almost all these methods neglect the outcome of the overfitting issue throughout the execution of DL techniques. This study presents a novel Enhancing Software-Defined Networking Security with Deep Learning and Hybrid Feature Selection (ESDNS-DLHFS) technique for consumer platforms. The proposed ESDNS-DLHFS system primarily focuses on protecting data privacy in SDN-assisted IoT platforms. In the ESDNS-DLHFS method, the initial phase of min-max normalization is executed to scale the input data. For the feature selection process, the hybrid crow search arithmetic optimization algorithm (HCSAOA) is utilized to optimally select feature subsets. Next, the deep bidirection- al long short-term memory (Deep BiL

关键词： Intrusion detection Feature extraction Software defined networking Computer science Information technology Deep learning classification algorithms Bidirectional long short term memory Telecommunication traffic Prediction algorithms Software-defined networking deep learning feature selection intrusion detection system enhanced artificial orcas algorithm

来源：评论

学校读者我要写书评

暂无评论

A Real-Time Intelligent System Based on Machine-Learning Methods for Improving Communication in Sign Language

引用

IEEE ACCESS 2025年 13卷 22055-22073页

作者： Leiva, Victor Rahman, Muhammad Zia Ur Akbar, Muhammad Azeem Castro, Cecilia Huerta, Mauricio Riaz, Muhammad Tanveer Pontificia Univ Catolica Valparaiso Sch Ind Engn Valparaiso 2340025 Chile Univ Engn & Technol Lahore Dept Mech Mechatron & Mfg Engn Faisalabad 38000 Pakistan Lappeenranta Univ Technol Dept Informat Technol Lappeenranta 53850 Finland Univ Minho Ctr Math P-4710057 Braga Portugal Univ Catolica Maule Dept Econ & Management Talca 3460000 Chile

In this article, we introduce a cost-effective and real-time intelligent system tailored to Pakistan sign language (PSL) recognition, aimed at facilitating communication for hearing-impaired individuals. The system utilizes a specialized glove equipped with flex sensors and an MPU-6050 device to capture finger movements and hand orientation in a three-dimensional space. A dataset comprising ten unique PSL signs, each performed by five participants for a total of 5000 samples, was used to train machine learning classifiers. These signs involve single-hand and single-movement gestures, optimizing the system for real-time PSL recognition. Machine learning classifiers, including decision trees, k-nearest neighbors, and support vector machines, achieved accuracy levels of 96%, 96.5%, and 97%, respectively. While direct quantitative comparisons with state-of-the-art systems are limited due to the uniqueness of PSL, we discuss our system in the context of recent advancements in sign language recognition. Real-time testing underscores the system's practical applicability and portability, demonstrating its potential for deployment in resource-constrained settings as an accessible initial step toward more comprehensive PSL recognition solutions.

关键词： Sign language Sensors Hands Support vector machines Nearest neighbor methods classification algorithms Real-time systems Vectors Training Flexible printed circuits Assistive technology flex sensor glove Kalman filter MPU-6050 device Raspberry Pi 3B microcontroller real-time processing sensor-based gesture recognition sign language recognition

来源：评论

学校读者我要写书评

暂无评论

Improving Sleep Disorder Diagnosis Through Optimized Machine Learning Approaches

引用

IEEE ACCESS 2025年 13卷 20989-21004页

作者： Rahman, Md. Atiqur Jahan, Israt Islam, Maheen Jabid, Taskeed Ali, Md Sawkat Rashid, Mohammad Rifat Ahmmad Islam, Mohammad Manzurul Ferdaus, Md. Hasanul Rasel, Md Mostofa Kamal Jahan, Mahmuda Rawnak Sharmin, Shayla Rimi, Tanzina Afroz Talukder, Atia Sanjida Matin, Md. Mafiul Hasan Ali, M. Ameer East West Univ Dept Comp Sci & Engn Dhaka 1212 Bangladesh Daffodil Int Univ Dept Comp Sci & Engn Daffodil Smart City 1216 Bangladesh Tesserent Melbourne Vic 3000 Australia

Classifying sleep disorders, such as obstructive sleep apnea and insomnia, is crucial for improving human quality of life due to their significant impact on health. The traditional expert-based classification of sleep stages, particularly through visual inspection, is challenging and prone to errors. This fact highlights the need for accurate machine learning algorithms (MLAs) for analyzing, monitoring, and diagnosing sleep disorders. This paper compares the MLAs for sleep disorder classification, specifically targeting None, Sleep Apnea, and Insomnia, using the Sleep Health and Lifestyle Dataset. We conducted two experiments. In the first one, we selected five key features from the feature spaces using the Gradient Boosting Regressor based on the Mean Decrease Impurity (MDI) technique. We chose two key features using the same methodology in the second experiment. We utilized 15 machine learning classifiers, and Gradient Boosting, Voting, Catboost, and Stacking Classifiers achieved an identical classification accuracy of 97.33%, with Precision, Recall, F1-score of 0.9733, and Specificity of 0.9569 in the original feature space. Among these, Gradient Boosting had the highest AUC of 0.9953 and was 3.36, 5.86, and 20.16 times faster than Voting, Catboost, and Stacking Classifiers, respectively. In the second experiment, the Decision Tree achieved the highest accuracy of 96% in the original and engineered feature spaces and was 149.33 times faster in the engineered feature space. Thus, this research proposes Gradient Boosting as the most effective method, outperforming all state-of-the-art techniques by achieving the highest accuracy, precision, recall, F1-score, and AUC, highlighting its superior classification performance and computational efficiency.

关键词： Sleep Sleep apnea Accuracy Brain modeling Machine learning Boosting Machine learning algorithms Data models Feature extraction classification algorithms SMOTEENN ANOVA test feature engineering classification sleep disorder machine learning classifiers ensemble technique

来源：评论

学校读者我要写书评

暂无评论

An Improved Bagging of Machine Learning algorithms to Predict Motif Structures From Protein-Protein Interaction Networks

引用

IEEE ACCESS 2025年 13卷 45077-45088页

作者： Lakshmi, P. Manikandan, P. Ramyachitra, D. Bharathiar Univ Dept Comp Sci Coimbatore 641046 Tamil Nadu India Vellore Inst Technol Sch Comp Sci & Engn Chennai 600127 Tamil Nadu India

Protein-protein interaction (PPI) networks play a vital role in the functional organization of cellular systems. Recurring subgraph patterns with these networks, known as motif structures, can be used to uncover mechanisms in complex biological processes. However, predicting motif structures from PPI networks remains challenging because of the complex and dynamic nature of interactions. Hence, this research focuses on predicting the topological structure of protein-protein interaction networks, acknowledging the computational challenges of exact motif discovery for diverse protein interactions. This study proposes a hybrid ensemble classification algorithm that leverages Penalized Matrix Decomposition (PMD) for feature extraction and a bagging approach using a Support Vector Machine (SVM), Alternative Decision Tree (ADT), and Extreme Learning Machine (ELM) classifiers. The proposed hybrid classification algorithm was evaluated against the ACC-Motif, Combi-Motif, Minimum Spanning Tree (MST), and Slider on benchmark datasets to demonstrate its statistical significance. On benchmark datasets, the proposed method achieved 92.3% accuracy, 91.5% precision, 89.8% recall, and an F1-score of 90.6%, outperforming existing algorithms such as ACC-Motif, Combi-Motif, and MST by 8-10%. This study assessed the practicality of similar topological subnetworks in protein-protein interaction networks, with Z-test calculations affirming superior outcomes. Overall, the proposed algorithm proved effective in predicting network structures, showing improved performance compared with other methods.

关键词： Proteins Matrix decomposition Feature extraction Prediction algorithms Topology classification algorithms Network topology Drug discovery Deep learning Motif penalized matrix decomposition (PMD) protein-protein interaction (PPI) bagging machine learning (ML) database of interacting proteins (DIP) database of interacting proteins (DIP)

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：