检索结果-内蒙古大学图书馆

An improved Artificial protozoa Optimizer for CNN architecture optimization

NEURAL NETWORKS 2025年 187卷 107368页

作者： Xie, Xiaofeng Gao, Yuelin Zhang, Yuming North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Ningxia Peoples R China Sci Comp & Intelligent Informat Proc Collaborat In Yinchuan 750021 Ningxia Peoples R China Ningxia Key Lab Intelligent Informat & Big Data Pr Yinchuan 750021 Ningxia Peoples R China

In this paper, we propose a novel neural architecture search (NAS) method called MAPOCNN, which leverages an enhanced version of the Artificial protozoa Optimizer (APO) to optimize the architecture of Convolutional Neural Networks (CNNs). The APO is known for its rapid convergence, high stability, and minimal parameter involvement. To further improve its performance, we introduce MAPO (Modified Artificial protozoa Optimizer), which incorporates the phototaxis behavior of protozoa. This addition helps mitigate the risk of premature convergence, allowing the algorithm to explore a broader range of possible CNN architectures and ultimately identify more optimal solutions. Through rigorous experimentation on benchmark datasets, including Rectangle and Mnist-random, we demonstrate that MAPOCNN not only achieves faster convergence times but also performs competitively when compared to other state-of-the-art NAS algorithms. The results highlight the effectiveness of MAPOCNN in efficiently discovering CNN architectures that outperform existing methods in terms of both speed and accuracy. This work presents a promising direction for optimizing deep learning architectures using biologically inspired optimization techniques.

关键词： Artificial protozoa Optimizer Neural architecture search Classification task MAPOCNN psoCNN

来源：评论

学校读者我要写书评

暂无评论

Large-scale machine learning with synchronous parallel adaptive stochastic variance reduction gradient descent for high-dimensional blindness detection on spark

引用

JOURNAL OF SUPERCOMPUTING 2025年第4期81卷 1-32页

作者： Qin, Chuandong Zhang, Yiqing Cao, Yu North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Peoples R China North Minzu Univ Ningxia Key Lab Intelligent Informat & big data Pr Yinchuan 750021 Peoples R China

The parallelization of optimization algorithms is of paramount importance in large-scale machine learning. In this paper, we explore the implementation of Adaptive learning rate Stochastic Gradient Descent (A-SGD) in a synchronized and parallelized manner. Additionally, we incorporate a Variance Reduction (VR) strategy to enhance the rate of convergence. Our approach addresses the complexity associated with high-dimensional datasets, particularly within the context of Logistic Regression (LR) and Support Vector Machine (SVM). Initially, we utilize the Histogram of Oriented Gradients (HOG) to extract high-dimensional sparse features from a dataset designed for Blindness Detection. Subsequently, we employ LR and SVM as our classifiers of choice. Finally, we apply the Synchronous A-SGD (SA-SGD) and Synchronous Adaptive Stochastic Variance Reduction (SA-SVRG) to the solutions of these classifiers. Our experimental results indicate that the performance of SA-SGD and SA-SVRG is notably superior when executed on a cluster as opposed to a single node.

关键词： Synchronous parallel algorithm High-dimensional sparse datasets Stochastic gradient descent Variance reduction Logistics regression Support vector machine

来源：评论

学校读者我要写书评

暂无评论

A stochastic recursive gradient algorithm integrating momentum and the powerball function with adaptive step sizes

引用

INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS 2025年 1-21页

作者： Qin, Chuandong Cai, Zilin Guo, Yuhang North Minzu Univ Sch Math & Informat Sci 204 Wenchang North St Yinchuan 750021 Ningxia Peoples R China North Minzu Univ Ningxia Key Lab Intelligent Informat & Big Data Pr 204 Wenchang North St Yinchuan 750021 Ningxia Peoples R China

Momentum techniques and the Powerball function have been proven effective in stochastic optimization algorithms, widely utilized in large-scale optimization scenarios. Nonetheless, the integration of these methodologies into stochastic optimization algorithms and the determination of their initial learning rates persist as unresolved and pivotal issues. In this study, we integrate momentum techniques and the Powerball function into the SARAH (StochAstic Recursive grAdient algoritHm), culminating in the inception of a novel variance-reduced gradient descent algorithm named PM-SARAH. Moreover, a pair of adaptive step size variants are respectively integrated into the outer and inner loops of PM-SARAH, giving rise to PM-SARAH-AS and PM-SARAH-RAS. Ultimately, through comparative experimentation with state-of-the-art optimization algorithms on standard machine learning tasks and certain non-convex scenarios, the empirical results underscore the superior performance of the algorithms elucidated in this paper.

关键词： Machine learning Variance reduction Momentum Powerball function Adaptive learning rate

来源：评论

学校读者我要写书评

暂无评论

A Multi-criteria Decision-Making Method Based on Intuitionistic Fuzzy Entropy and Three-Way Ranking TOPSIS with Applications

引用

INTERNATIONAL JOURNAL OF FUZZY SYSTEMS 2025年 1-24页

作者： Xu, Changlin Lan, Yushi North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Ningxia Peoples R China North Minzu Univ Key Lab Intelligent Informat & Big Data Proc Ningx Yinchuan 750021 Ningxia Peoples R China

This article presents a comprehensive analysis of the current developmental status of the intuitionistic fuzzy entropy method (IFEM). Initially, it introduces the foundational theory of intuitionistic fuzzy sets (IFSs) and the IFEM. By introducing the intuitionistic fuzzy distance measure (IFDM) to assess the similarity between IFSs, understanding their fuzziness can be enhanced. Subsequently, an analysis of numerical examples evaluates both the advantages and disadvantages of existing intuitionistic fuzzy entropy methods (IFEMs). To address the inconsistencies in the existing axiomatic definition of IFEM, this paper optimizes the definition to comprehensively account for the influences of the deviation of membership degree (MD) and non-membership degree (NMD), as well as hesitation degree (HD). These optimized axiomatic definitions improve the accuracy of uncertainty measurement for IFSs. Secondly, an IFEM based on enhanced axiomatic conditions is proposed, integrating the characteristics of the cosine, exponential, and logarithmic functions. Its validity is verified through mathematical reasoning. This paper demonstrates the rationality of the proposed method by simulating the analysis and comparison of ranking results for IFSs metrics, using existing IFEMs alongside the newly introduced method. Finally, this paper constructs a multi-criteria decision-making (MCDM) model based on the IFEM introduced herein, integrating intuitionistic fuzzy Three-Way Ranking (TWR) and Technique for Order preference by Similarity to Ideal Solution (TOPSIS) concepts to address the medical decision-making problem of selecting nanomaterials for bone transplant surgery in amputee patients. The feasibility of the proposed method is further validated through comparative analysis.

关键词： Intuitionistic fuzzy entropy Axiomatic definition Three-way ranking Multi-criteria decision-making

来源：评论

学校读者我要写书评

暂无评论

Application of Wiener Filter Based on Improved BB Gradient Descent in Iris Image Restoration

引用

JOURNAL OF IMAGING informatICS IN MEDICINE 2025年第2期38卷 1165-1183页

作者： Qin, Chuandong Zhang, Yiqing North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Peoples R China Ningxia Key Lab Intelligent Informat & Big Data Pr Yinchuan 750021 Peoples R China

Iris recognition, renowned for its exceptional precision, has been extensively utilized across diverse industries. However, the presence of noise and blur frequently compromises the quality of iris images, thereby adversely affecting recognition accuracy. In this research, we have refined the traditional Wiener filter image restoration technique by integrating it with a gradient descent strategy, specifically employing the Barzilai-Borwein (BB) step size selection. This innovative approach is designed to enhance both the precision and resilience of iris recognition systems. The BB gradient method is adept at optimizing the parameters of the Wiener filter by introducing simulated blurring and noise conditions to the iris images. Through this process, it is capable of restoring images that have been degraded by blur and noise, leading to a significant improvement in the clarity of the restored images and, consequently, a notable elevation in recognition performance. The results of our experiments have demonstrated that this advanced method surpasses conventional filtering techniques in terms of both subjective visual quality assessments and objective peak signal-to-noise ratio (PSNR) evaluations.

关键词： Iris restoration Barzilai-Borwein step size Gradient descent Wiener filter

来源：评论

学校读者我要写书评

暂无评论

probability Density Evolution and Fractional PID Control of Magnetic Bearing-Rigid Rotor System Influenced by Multisource Stochastic Factors

引用

JOURNAL OF NONLINEAR MATHEMATICAL PHYSICS 2025年第1期32卷 1-20页

作者： Ma, Shaojuan Wu, Maixia Ma, Xiaoyan Liu, Qiyi North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Peoples R China Ningxia Key Lab Intelligent Informat & Big Data Pr Yinchuan 750021 Peoples R China Ningxia Basic Sci Res Ctr Math Yinchuan 750021 Peoples R China

Magnetic bearings are widely used in engineering for their remarkable advantages of low friction, no wear and high speed. In this paper, the dynamics and control of the magnetic bearing-rigid rotor system under the influence of multisource stochastic factors are studied. Firstly, a model of magnetic bearing-rigid rotor system driven by multiplicative noise and additive noise is established. Secondly, the stationary response probability density function (PDF) of the proposed magnetic bearing system is obtained by using the amplitude envelope stochastic average method. Then we analyze the influence of different system parameters and noise intensity on the magnetic bearing-rigid rotor system, and find that internal stochastic factors are the main factors affecting the stability of the bearing system. Finally, the fractional proportional Integral Derivative (PID) control of the system is studied. Based on the numerical analysis, we verify the effectiveness of the control strategy and draw the conclusion that adjusting appropriately the parameters of fractional PID controller can make the magnetic bearing system control better.

关键词： Magnetic bearing-rigid rotor system Multisource stochastic factors Stochastic average method Fractional PID control

来源：评论

学校读者我要写书评

暂无评论

Parameter identification of the Black-Scholes model driven by multiplicative fractional Brownian motion

引用

PHYSICA A-STATISTICAL MECHANICS AND ITS APPLICATIONS 2025年 660卷

作者： Hou, Wentao Ma, Shaojuan North Minzu Univ Sch Math & Informat Sci Yinchuan 750021 Ningxia Peoples R China Yuncheng Univ Dept Math & Informat Technol Yuncheng 044000 Shanxi Peoples R China Key Lab Intelligent Informat & Big Data Proc Ning Yinchuan 750021 Ningxia Peoples R China

In this paper, we propose a parameter identification method based on deep learning network, which can jointly identify all parameters of the Black-Scholes (BS) model driven by multiplicative fractional Brownian motion (FBM) in a discrete sample trajectory. Firstly, the Convolutional Neural Network (CNN) is combined with the Bi-directional Gated Recurrent Unit (bigRU) and the attention mechanism (AM) is integrated to construct the new identifier (CBANN). Then, the multiplicative FBM is constructed as the random effect of the BS model, and all the parameters of the model are identified by the new identifier. Finally, extensive numerical simulations are conducted for both known and unknown Hurst exponents, and two empirical studies are performed using real data. The results suggest that, compared to the PENN identifier and the maximum likelihood (ML) identifier, the proposed identifier can simultaneously identify all parameters in the model more quickly and accurately. Additionally, several advantages of the new identifier are discussed, including its strong generalization performance, flexibility in training set proportion settings, and the incorporation of an attention mechanism layer.

关键词： Dynamical system Fractional brownian motion Deep learning network Parameter identification

来源：评论

学校读者我要写书评

暂无评论

Stochastic Optimal Control of Multi-Machine Power Systems Driven by Fractional Gaussian Noise

引用

INTERNATIONAL JOURNAL OF ROBUST AND NONLINEAR CONTROL 2025年

作者： Zhang, Chengjian Ma, Shaojuan Li, Hufei Xiao, Hui North Minzu Univ Sch Math & Informat Sci Yinchuan Peoples R China North Minzu Univ Ningxia Key Lab Intelligent Informat & Big Data Pr Yinchuan Peoples R China Shanghai Univ Sch Mech & Engn Sci Shanghai Peoples R China Shanghai Univ Shanghai Inst Appl Math & Mech Shanghai Peoples R China

This paper studies the optimal control of multi-machine power systems with the excitation of fractional Gaussian noise (FGN) and applies the theory of quasi-generalized Hamiltonian systems. First, a multi-machine power system model is established based on the long-range correlation of fractional Gaussian noise, which is transformed into a quasi-generalized Hamiltonian system model. Next, according to the fractional path integral rules and the stochastic averaging method (SAM), the stochastic multi-machine power system is reduced in dimension. Then, based on the averaged multi-machine power system, the dynamic programming method of stochastic optimal control is utilized to propose a control strategy that maximizes the probability of the controlled system being in the desired region. Finally, the effectiveness of the proposed control strategy is verified through numerical simulations, and the superiority of the control methods has also been compared with the Linear Quadratic Regulator (LQR).

关键词： fractional Gaussian noise intra-region probability maximization multi-machine power systems quasi-generalized Hamiltonian system stochastic optimal control

来源：评论

学校读者我要写书评

暂无评论

Joint resource allocation and privacy protection for MEC task offloading in industrial Internet

引用

CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS 2025年第2期28卷 1-26页

作者： Wang, Zhongmin Zhang, Shuai Jin, Xiaomin Chen, Yanping Lu, Chen Xian Univ Posts & Telecommun Sch Comp Sci & Technol Xian 710121 Shaanxi Peoples R China Shaanxi Key Lab Network Data Anal & Intelligent Pr Xian 710121 Shaanxi Peoples R China Xian Key Lab Big Data & Intelligent Comp Xian 710121 Shaanxi Peoples R China Shenzhen Inst Informat Technol Shenzhen 518172 Guangdong Peoples R China

The rapid development of the industrial Internet has led to significant advancements in manufacturing information. To address the limitations of traditional cloud computing architectures, such as high latency, inefficiency, and reduced flexibility, multi-access edge computing (MEC) has emerged as a promising paradigm. By enabling task offloading to edge or cloud servers, MEC can significantly reduce the task response latency and mitigate data privacy risk. However, despite the potential of MEC, many existing solutions still face challenges in balancing latency, resource allocation, and privacy protection in real-world scenarios. To address these challenges, this paper establishes a novel joint resource allocation and privacy protection offloading optimization model for multi-network transmission tasks. The model integrates offloading ratio, resource allocation, and network selection into a unified framework to minimize latency while addressing critical factors such as terminal security, latency sensitivity, and resource constraints. We formulate this problem and prove its NP-hardness before decomposing it into two subproblems. The upper layer solves the optimal offloading strategy, while the lower layer determines a reasonable resource allocation strategy based on the optimal upper-layer solution. In response, we propose a dual-layer offloading algorithm based on an adaptive parameter adjustment of the dung beetle optimizer with Bernoulli mapping. Experimental results demonstrate that our offloading model effectively enhances privacy protection, reduces latency, and optimizes overall costs. Compared with existing offloading algorithms, our algorithm shows superior convergence and offloading performance, reducing weighted costs by at least 16.91%.

关键词： Industrial Internet Multi-access edge computing Task offloading privacy protection Dual-layer offloading algorithm

来源：评论

学校读者我要写书评

暂无评论

An Efficient Illumination-Invariant Dynamic Facial Expression Recognition for Driving Scenarios

引用

IET intelligent TRANSPORT SYSTEMS 2025年第1期19卷

作者： Pei, Ercheng Guo, Man Berenguer, Abel Diaz He, Lang Chen, Haifeng Xian Univ Posts & Telecommun Sch Comp Sci & Technol Xian Shaanxi Peoples R China Shaanxi Key Lab Network Data Anal & Intelligent Pr Xian Peoples R China Xian Key Lab Big Data & Intelligent Comp Xian Shaanxi Peoples R China Vrije Univ Brussel Dept Elect & Informat ETRO Brussels Belgium Shaanxi Univ Sci & Technol Sch Elect Informat & Artificial Intelligence Xian Shaanxi Peoples R China

Facial expression recognition (FER) is significant in many application scenarios, such as driving scenarios with very different lighting conditions between day and night. Existing methods primarily focus on eliminating the negative effects of pose and identity information on FER, but overlook the challenges posed by lighting variations. So, this work proposes an efficient illumination-invariant dynamic FER method. To augment the robustness of FER methods to illumination variance, contrast normalisation is introduced to form a low-level illumination-invariant expression features learningmodule. In addition, to extract dynamic and salient expression features, a two-stage temporal attention mechanism is introduced to form a high-level dynamic salient expression features learning module deciphering dynamic facial expression patterns. Furthermore, additive angular margin loss is incorporated into the training of the proposed model to increase the distances between samples of different categories while reducing the distances between samples belonging to the same category. We conducted comprehensive experiments using the Oulu-CASIA and DFEW datasets. On the Oulu-CASIA VIS and NIR subsets in the normal illumination, the proposed method achieved accuracies of 92.08% and 91.46%, which are 1.01 and 7.06 percentage points higher than the SOTA results from the DCBLSTM and CELDL method, respectively. Based on the Oulu-CASIA NIR subset in the dark illumination, the proposed method achieved an accuracies of 91.25%, which are 4.54 percentage points higher than the SOTA result from the CDLLNet method. On the DFEW dataset, the proposed method achieved a UAR of 60.67% and a WAR of 71.48% with 12M parameters, approaching the SOTA result from the VideoMAE model with 86M parameters. The outcomes of our experiments validate the effectiveness of the proposed dynamic FER method, affirming its ability in addressing the challenges posed by diverse illumination conditions in driving scenarios.

关键词： driving scenarios dynamic facial expression recognition illumination invariance

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：