检索结果-内蒙古大学图书馆

Inter-class margin climbing with cost-sensitive learning in neural network classification

KNOWLEDGE AND INFORMATION SYSTEMS 2025年第2期67卷 1993-2016页

作者： Zhang, Siyuan Xie, Linbo Chen, Ying Zhang, Shanxin Jiangnan Univ Sch Internet Things Engn 1800 Lihu Ave Wuxi 214122 Jiangsu Peoples R China

Large margin stands as an intuitive indicator of reliable classifiers, reflecting classifier robustness and generalizability. However, due to the intricate nonlinear mapping, directly defining margin as the optimization objective for multi-layer neural network is always challenging. On the other hand, the cost sensitivity coefficients of individual samples hold promise for shaping decision boundaries of neural network classification. A question arises as to whether optimizing neural network classifier can be guided to achieve larger classification margin by varying instance-level cost sensitivity factor. Inspired by above question, this paper proposes a heuristic hard mining strategy designed to progressively identify challenging samples and amplify the output margin through cost-sensitive learning. The refinement process adjusts the sample distribution when optimization reaches a local minimum to ensure the sustainable optimization, ultimately leading to margin climbing. Two hard mining algorithms are designed for binary and multi-class classification problems, which utilize distinct margin definitions based on different decision-making scenarios. In the proposed method, we focus on establishing individualized margin between distinct categories to more accurately characterize the inter-class margin. Empirical results demonstrate that our proposed methodology enhances both the accuracy and robustness in neural network classification.

关键词： Margin maximization Cost-sensitive learning Hard mining strategy neural network classification

来源：评论

学校读者我要写书评

暂无评论

Statistical analysis of aging micro-characteristics for asphalt binder AFM images based on neural network classification

引用

FUEL 2024年第PartB期371卷

作者： Yu, Tengjiang Wang, Jingjing Zhang, Haitao Sun, Junzu Ludong Univ Sch Civil Engn Yantai 264025 Peoples R China Weifang Univ Sci & Technol Sch Architectural Engn Weifang 262700 Peoples R China Northeast Forestry Univ Coll Civil Engn Harbin 150040 Peoples R China

To further evaluate the effect of aging behavior on microstructure for asphalt binder, the atomic force microscope (AFM) images of asphalt binder were studied by neural network classification technology, and its aging micro-characteristics were analyzed statistically. Firstly, AFM equipment was used to test 90# virgin asphalt binder and styrene-butadiene-styrene (SBS) modified asphalt binder commonly used in engineering, and the typical micromorphology AFM images of asphalt binders were obtained. Subsequently, the asphalt binder AFM images were preliminatively analyzed, and the micro-characteristics of AFM images before and after aging were compared, including the "bee structure" characteristics, distribution law and surface structure roughness. Finally, the asphalt binder AFM images were further analyzed by neural network classification, and the change of "bee structure" was statistically calculated and analyzed, and its aging micro-characteristics were quantitatively evaluated. The results show that it is feasible for neural network classification to be applied to the statistical analysis of asphalt binder AFM images, and it can further describe and evaluate the aging micro-characteristics of asphalt binder. Aging behavior reduces the proportion of intermediate category areas in asphalt binder AFM images and increases the "bee structure" areas. The cumulative percentage curve of asphalt binder AFM images is smoother (each category is more uniform), indicating that the asphalt binder performance tended to be stable after aging. The results further enrich the description of asphalt binder AFM images, and provide a new way to explore the micro-morphology of asphalt binder.

关键词： Asphalt binder AFM images neural network classification Aging micro -characteristics Statistical analysis

来源：评论

学校读者我要写书评

暂无评论

Leader learning loss function in neural network classification

引用

NEUROCOMPUTING 2023年第1期557卷

作者： Zhang, Siyuan Xie, Linbo Jiangnan Univ Sch Internet Things Engn Wuxi Jiangsu Peoples R China Minist Educ Internet Things Technol Applicat Engn Res Ctr Wuxi Jiangsu Peoples R China

Deep learning, based on Empirical Risk Minimization (ERM), typically aims to fit the ideal outputs of all samples due to its large capacity. However, models trained based on empirical losses like cross entropy (CE) or mean square error (MSE), often learn unnecessary information during classification, leading to premature overfitting. On the other hand, the result-focused loss functions, i.e., zero-one loss or hinge loss, are hard to optimize and thus are rarely applied directly in neural network. This paper proposes a novel leader learning in classification, where CE is gradually trained by classification results using sample-dependent cost-sensitive learning. As complementary, the stepwise-changed CE covers the deficiency on classification error while preserving the advantage of fast convergence. In this way, the deviation between CE and classification error can be corrected. Experimental results demonstrate that the proposed leader learning has a more significant convergence trend than the baseline algorithms. Moreover, the loss function learned from a specific dataset has broad generality that can be transferred to other models as prior knowledge.

关键词： Leader learning Loss function learning neural network classification Cost-sensitive learning

来源：评论

学校读者我要写书评

暂无评论

MPCE: A Maximum Probability Based Cross Entropy Loss Function for neural network classification

引用

IEEE ACCESS 2019年 7卷 146331-146341页

作者： Zhou, Yangfan Wang, Xin Zhang, Mingchuan Zhu, Junlong Zheng, Ruijuan Wu, Qingtao Henan Univ Sci & Technol Coll Informat Engn Luoyang 471023 Peoples R China Shanghai Int Studies Univ Lab Appl Brain & Cognit Sci Postdoctoral Res Stn Sch Business & Management Shanghai 200083 Peoples R China Henan Qunzhi Informat Technol Co Ltd Luoyang 471003 Peoples R China Guangzhou Xiangxue Pharmaceut Co Ltd Guangzhou 510663 Guangdong Peoples R China

In recent years, multi-classifier learning is of significant interest in industrial and economic fields. Moreover, neural network is a popular approach in multi-classifier learning. However, the accuracies of neural networks are often limited by their loss functions. For this reason, we design a novel cross entropy loss function, named MPCE, which based on the maximum probability in predictive results. In this paper, we first analyze the difference of gradients between MPCE and the cross entropy loss function. Then, we propose the gradient update algorithm based on MPCE. In the experimental part of this paper, we utilize four groups of experiments to verify the performance of the proposed algorithm on six public datasets. The first group of experimental results show that the proposed algorithm converge faster than the algorithms based on other loss functions. Moreover, the results of the second group show that the proposed algorithm obtains the highest training and test accuracy on the six datasets, and the proposed algorithm perform better than others when class number changing on the sensor dataset. Furthermore, we use the model of convolutional neural network to implement the compared methods on the mnist dataset in the fourth group of experiments. The results show that the proposed algorithm has the highest accuracy among all executed methods.

关键词： Cross entropy loss function maximum probability neural network classification softmax

来源：评论

学校读者我要写书评

暂无评论

neural network based time zone classification allows circadian rhythm analysis 60

Neural network based time zone classification allows circadi...

引用

60th Annual Conference of the Society-of-Instrument-and-Control-Engineers-of-Japan (SICE)

作者： Shigemoto, Masaya Natsume, Kiyohisa Kyushu Inst Technol Dept Human Intelligence Syst Fukuoka Japan

ISBN: (纸本)9784907764739

The circadian rhythm modulates memory in animals. However, the relationship between the circadian cycle and brain wave dynamics has not yet been clarified. Beta wave-like oscillations with frequencies in the range of 13-30 Hz can be induced by applying a cholinergic agonist, carbachol (CCh). We estimated the time of day using a neural network with oscillatory parameters. Rats were kept under light/dark (LD) condition or dark/dark (DD) condition, and hippocampal slices were used for analysis. CCh-induced beta oscillation (CIBO) was induced with CCh at all time zones in the four-hour bins. CIBO showed significant differences in duration, and inter-burst interval (IBI). Additionally, there were significant differences between LD and DD conditions within the same time zones. CIBO duration, IBI and amplitude were not affected by GABAA receptor antagonist (SR95531) during the light phase, which is different from our previous results on CIBO frequency. Finally, neural network classifiers that estimated six time zones using CIBO parameters were developed and evaluated. We obtained an average of accuracy rate of 78%. Our results suggest that the brain wave changes depend on the time of day, and the inhibitory neural system may be involved in the change partially. Further, we showed that it is possible to develop time zone classifiers using brain waves.

关键词： Hippocampus Brain wave Circadian rhythm neural network classification

来源：评论

学校读者我要写书评

暂无评论

Rapid optimal control law generation: an MoE based method

引用

Journal of Systems Engineering and Electronics 2025年第1期36卷 280-291页

作者： ZHANG Tengfei SU Hua GONG Chunlin YANG Sizhi BAI Shaobo Shaanxi Aerospace Flight Vehicle Design Key Laboratory School of AstronauticsNorthwestern Polytechnical UniversityXi’an 710072China Northwest Industries Group Company Xi’an 710043China Northwest Institute of Mechanical and Electrical Engineering Xianyang 712099China

To better complete various missions, it is necessary to plan an optimal trajectory or provide the optimal control law for the multirole missile according to the actual situation, including launch conditions and target location. Since trajectory optimization struggles to meet real-time requirements, the emergence of data-based generation methods has become a significant focus in contemporary research. However, due to the large differences in the characteristics of the optimal control laws caused by the diversity of tasks, it is difficult to achieve good prediction results by modeling all data with one single ***, the modeling idea of the mixture of experts(MoE) is adopted. Firstly, the K-means clustering algorithm is used to partition the sample data set, and the corresponding neural network classification model is established as the gate switch of MoE. Then, the expert models, i.e., the mappings from the generation conditions to the optimal control law represented by the results of principal component analysis(PCA), are represented by Kriging models. Finally, multiple rounds of accuracy evaluation, sample supplementation, and model updating are conducted to improve the generation accuracy. The Monte Carlo simulation shows that the accuracy of the proposed model reaches 96% and the generation efficiency meets the real-time requirement.

关键词： optimal control mixture of experts(MoE) K-means Kriging model neural network classification principal component analysis(PCA)

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral detection of blood and melanin defects in salmon fillets

引用

AQUACULTURE 2025年 602卷

作者： Romeyn, Rowan Ortega, Samuel Heia, Karsten Nofima AS NOFIMA Tromso Postboks 6122 NO-9291 Tromso Norway

This study investigates the use of hyperspectral imaging (HSI) and neural network classification for detecting blood and melanin spots in salmon fillets. Hyperspectral images were collected and categorized into training (100 fillets), cross-validation (28 fillets), and test (229 fillets) groups to develop and evaluate a neural network classification model. Model classification performance was assessed at both pixel-level and at the object level by using a connected component analysis to detect connected regions in the classification maps. Blood spots were detected with higher recall (0.81 at pixel level and 0.92 at object level) than melanin spots (0.71 at pixel level and 0.86 at object level) in the holdout cross validation set, likely due to the distinctive absorption spectrum of hemoglobin compared to eumelanin. The corresponding F1 score, reflecting the harmonic mean of recall and precision, was 0.76 for blood and 0.68 for melanin at the pixel level. The F1 scores at the object level were higher, with scores of 0.85 for blood and 0.84 for melanin, since mismatch in the exact boundary of the defects affects the pixel level scores more than the object level scores. In general, the model effectively separated between blood and melanin defect types, although transient states between pure blood and pure melanin were also observed, particularly in the predicted class probabilities, in line with observations made by previous studies. Directly analyzing the modelled probabilities of blood and melanin may better resolve these transient states, rather than simply classifying the most probable class. The model maintained robust performance across different test sets, with an F1 score of 0.86 achieved for melanin spot detection in an image set (120 fillets) acquired at a different facility, indicating its ability to generalize to new data. Spatial registration of all images with a single reference fillet allowed a common spatial reference to be established. Spatial mapping

关键词： Hyperspectral imaging neural network classification Melanin spots Blood spots Salmon fillets Defect detection Quality control

来源：评论

学校读者我要写书评

暂无评论

SPARSE CODING BASED FEATURE EXTRACTION TECHNIQUEFOR PERSON IDENTIFICATION

引用

INTERNATIONAL JOURNAL OF INNOVATIVE COMPUTING INFORMATION AND CONTROL 2024年第4期20卷 1077-1091页

作者： Samkunta, Jirayu Ketthong, Patinya Mai, Nghia Thi Hashikura, Kotaro Kamal, Md Abdus Samad Murakami, Iwanori Yamada, Kou Gunma Univ Grad Sch Sci & Technol 1-5-1 Tenjincho Kiryu 3768515 Japan Gunma Univ Div Mech Sci & Technol 1-5-1 Tenjincho Kiryu 3768515 Japan Thai Nichi Inst Technol Fac Engn 1771-1 Pattanakarn Rd Suanluang 10250 Bangkok Thailand Posts & Telecommun Inst Technol Dept Elect & Elect 1 Km 10 Hanoi 151090 Vietnam

Human identification research has been an important topic invarious fields,ranging from biometrics to human-computer interaction. However, extracting meaningfulinformation from hand movements to facilitate accurate personal identification remainsa significant challenge. This study addresses this challenge by proposing an alternativeapproach for feature extraction of hand kinematics using a sparse coding technique. Asparse coding technique is utilized to extract hand motion features from time series da-ta, enabling personalized identification of individuals. Despite numerous existing featureextraction techniques, sparse coding offers distinct advantages, including the ability to re-duce the number of features while retaining essential information. The proposed methodis evaluated using hand kinematics time series data, and itseffectiveness is assessedthrough the neural network (NN) classification technique. Experimental validations areconducted to demonstrate and compare the effectiveness of the proposed method againststate-of-the-art sparse coding and other feature extraction techniques. The results revealthat the proposed method outperforms existing techniques,achieving significantly high-er accuracy in participant identification. Overall, this study presents a novel approachto feature extraction for hand kinematics-based human classification, offering promisingimplications for various applications, including biometrics and human-computer interaction.

关键词： Hand kinematics Feature extraction Sparse coding neural network classification Hand kinematics time series Human recognition

来源：评论

学校读者我要写书评

暂无评论

The Combination of Topological Data Analysis and Mathematical Modeling Improves Sleep Stage Prediction From Consumer-Grade Wearables

引用

JOURNAL OF BIOLOGICAL RHYTHMS 2024年第6期39卷 535-553页

作者： Lee, Minki P. Kim, Dae Wook Mayer, Caleb Walch, Olivia Forger, Daniel B. Univ Michigan Dept Math Ann Arbor MI USA Korea Adv Inst Sci & Technol KAIST Dept Brain Cognit Sci Daejeon 34141 South Korea Sogang Univ Dept Math Seoul South Korea Stanford Univ Dept Genet Stanford CA USA Arcascope Arlington VA USA Univ Michigan Dept Neurol Ann Arbor MI USA Univ Michigan Dept Computat Med & Bioinformat Ann Arbor MI USA Univ Michigan Michigan Ctr Appl & Interdisciplinary Math Ann Arbor MI USA

Wearable devices have become commonplace tools for tracking behavioral and physiological parameters in real-world settings. Nonetheless, the practical utility of these data for clinical and research applications, such as sleep analysis, is hindered by their noisy, large-scale, and multidimensional characteristics. Here, we develop a neural network algorithm that predicts sleep stages by tracking topological features (TFs) of wearable data and model-driven clock proxies (CPs) reflecting the circadian propensity for sleep. To evaluate its accuracy, we apply it to motion and heart rate data from the Apple Watch worn by young subjects undergoing polysomnography (PSG) and compare the predicted sleep stages with the corresponding ground truth PSG records. The neural network that includes TFs and CPs along with raw wearable data as inputs shows improved performance in classifying Wake/REM/NREM sleep. For example, it shows significant improvements in identifying REM and NREM sleep (AUROC/AUPRC improvements >13% and REM/NREM accuracy improvement of 12%) compared with the neural network using only raw data inputs. We find that this improvement is mainly attributed to the heart rate TFs. To further validate our algorithm on a different population, we test it on elderly subjects from the Multi-ethnic Study of Atherosclerosis cohort. This confirms that TFs and CPs contribute to the improvements in Wake/REM/NREM classification. We next compare the performance of our algorithm with previous state-of-the-art wearable-based sleep scoring algorithms and find that our algorithm outperforms them within and across different populations. This study demonstrates the benefits of combining topological data analysis and mathematical modeling to extract hidden inputs of neural networks from puzzling wearable data.

关键词： sleep scoring wearables heart rate motion topological data analysis mathematical modeling neural network classification

来源：评论

学校读者我要写书评

暂无评论

Multimodal neural network for Recognition of Cardiac Arrhythmias Based on 12-Load Electrocardiogram Signals

引用

IEEE ACCESS 2023年 11卷 133744-133754页

作者： Kiladze, Mariya R. Lyakhova, Ulyana A. Lyakhov, Pavel A. Nagornov, Nikolay N. Vahabi, Mohsen North Caucasus Fed Univ Dept Math Modeling Stavropol 355017 Russia North Caucasus Fed Univ North Caucasus Ctr Math Res Stavropol 355017 Russia Shahrood Univ Technol Fac Elect Engn Shahrud *** Iran

Automatic classification of heart rhythm disturbances using an electrocardiogram is a reliable way to timely detect diseases of the cardiovascular system. The need to automate this process is to increase the number of electrocardiogram signals. classification methods based on the use of neural networks provide a high percentage of arrhythmia recognition. However, known classification methods do not take into account patient characteristics. The work proposes a multimodal neural network that takes into account the age and gender characteristics of the patient. It includes a Long short-term memory (LSTM) network for feature extraction on twelve-channel electrocardiogram signals and a linear neural network for processing patient metadata such as age and gender. Extraction of electrocardiogram signal features occurs in parallel with metadata processing. The last unifying layer of the proposed multimodal neural network integrates heterogeneous data and features of electrocardiogram signals obtained using an LSTM network. The developed multimodal neural network was verified using the PhysioNet/Computing in Cardiology Challenge 2021 ECG database. The simulation results showed that the proposed multimodal neural network achieves a recognition accuracy of 63%, which is 2 percentage points higher compared to state-of-the-art methods.

关键词： neural network classification metadata linear perceptron LSTM network PhysioNet/Computing in Cardiology Challenge 2021

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：