检索结果-内蒙古大学图书馆

Q-ADER: An Effective Q-Learning for Recommendation With Diminishing Action Space

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS 2025年第5期36卷 8510-8524页

作者： Li, Fan Qu, Hong Zhang, Liyan Fu, Mingsheng Chen, Wenyu Yi, Zhang Sichuan Univ West China Univ Hosp 2 Dept Obstet & Gynecol Key Lab Birth Defects & Related Dis Women & Childr Chengdu 610041 Peoples R China Univ Elect Sci & Technol China Sch Comp Sci & Engn Chengdu 611731 Peoples R China State Grid Sichuan Elect Power Co Mkt Serv Ctr Chengdu 610041 Peoples R China Sichuan Univ Sch Comp Sci Chengdu 610065 Peoples R China

Deep reinforcement learning (RL) has been widely applied to personalized recommender systems (PRSs) as they can capture user preferences progressively. Among RL-based techniques, deep Q-network (DQN) stands out as the most popular choice due to its simple update strategy and superior performance. Typically, many recommendation scenarios are accompanied by the diminishing action space setting, where the available action space will gradually decrease to avoid recommending duplicate items. However, existing DQN-based recommender systems inherently grapple with a discrepancy between the fixed full action space inherent in the Q-network and the diminishing available action space during recommendation. This article elucidates how this discrepancy induces an issue termed action diminishing error in the vanilla temporal difference (TD) operator. Due to this discrepancy, standard DQN methods prove impractical for learning accurate value estimates, rendering them ineffective in the context of diminishing action space. To mitigate this issue, we propose the Q-learning-based action diminishing error reduction (Q-ADER) algorithm to modify the value estimate error at each step. In practice, Q-ADER augments the standard TD learning with an error reduction term which is straightforward to implement on top of the existing DQN algorithms. Experiments are conducted on four real-world datasets to verify the effectiveness of our proposed algorithm.

关键词： Recommender systems Standards Q-learning Training Temporal difference learning Reliability prediction algorithms Error reduction recommender systems value estimate error

来源：评论

学校读者我要写书评

暂无评论

Model-Free Predictive Control Based on RLS Algorithm for Grid-Forming Inverters With Virtual Synchronous Generator

引用

IEEE ACCESS 2025年 13卷 57921-57931页

作者： Huang, Min Zhang, Huiying Wang, Kangan Yao, Zhilei Wu, Weimin Shanghai Maritime Univ Dept Elect Engn Shanghai 201306 Peoples R China

As a key element to connect renewable energy sources, grid-forming inverters (GFIs) mostly use droop control as an outer loop, but it can cause a swift change in the system frequency when disturbances arise. In the role of the inner loop control, the traditional model predictive control (MPC) is widely employed in converters, but it is heavily dependent on model parameters. To overcome the above drawbacks, a model-free predictive control with virtual synchronous generator (VSG-MFPC) is proposed in this study. The scheme uses an autoregressive MFPC based on an auto-regressive with exogenous input (ARX) model as an inner-loop control structure for voltage prediction, combined with a simplified virtual synchronous generator (VSG) as an outer-loop control structure for power sharing and inertial support control systems. The implementation of MFPC improves the parameter robustness of the system. Building on this, the improvement schemes of algorithmic parameter initialization, delay compensation and current limiting capability are further investigated. Additionally, incorporating the VSG into the outer loop strengthens the stability of the system's frequency and voltage while also enhancing the system's dynamic characteristics. The aforementioned advantages, along with the efficacy and feasibility of the proposed control method, are validated through simulation and experimental results.

关键词： Predictive models Voltage control Predictive control Microgrids Synchronous generators Computational modeling Adaptation models Switches prediction algorithms Complexity theory Grid forming inverters model predictive control model free predictive control virtual synchronous generator

来源：评论

学校读者我要写书评

暂无评论

Memory-Enhanced Distributed Accelerated algorithms for Coordinated Linear Computation

引用

IEEE TRANSACTIONS ON SIGNAL AND INFORMATION PROCESSING OVER NETWORKS 2025年 11卷 35-47页

作者： Ding, Shufen Meng, Deyuan Du, Mingjun Cai, Kaiquan Beihang Univ BUAA Sch Automat Sci & Elect Engn Beijing 100191 Peoples R China Beihang Univ BUAA Res Div 7 Beijing 100191 Peoples R China Beihang Univ BUAA State Key Lab CNS ATM Beijing 100191 Peoples R China Qilu Univ Technol Shandong Acad Sci Sch Informat & Automat Engn Jinan 250353 Peoples R China Beihang Univ BUAA Sch Elect & Informat Engn Beijing 100191 Peoples R China

In this paper, a memory-enhanced distributed accelerated algorithm is proposed for solving large-scale systems of linear equations within the context of multi-agent systems. By employing a local predictor consisting of a linear combination of the nodes' current and previous values, the inclusion of two memory taps can be characterized such that the convergence of the distributed solution algorithm for coordinated computation is accelerated. Moreover, consensus-based convergence results are established by leveraging an analysis of the spectral radius of an augmented iterative matrix associated with the error system that arises from solving the equation. In addition, the connection between the convergence rate and the tunable parameters is developed through an examination of the spectral radius of the iterative matrix, and the optimal mixing parameter is systematically derived to achieve the fastest convergence rate. It is shown that despite whether the linear equation of interest possesses a unique solution or multiple solutions, the proposed distributed algorithm exhibits exponential convergence to the solution, without dependence on the initial conditions. In particular, both the theoretical analysis and simulation examples demonstrate that the proposed distributed algorithm can achieve a faster convergence rate than conventional distributed algorithms for the coordinated linear computation, provided that adjustable parameters are appropriately selected.

关键词： Convergence Distributed algorithms Signal processing algorithms prediction algorithms Vectors Jacobian matrices Symmetric matrices Multi-agent systems Information processing Distributed databases Convergence rate consensus distributed algorithms iterative method multi-agent systems

来源：评论

学校读者我要写书评

暂无评论

Ultra-Short-Term Wind Power Forecasting Based on DT-DSCTransformer Model

引用

IEEE ACCESS 2025年 13卷 22919-22930页

作者： Gao, Yanlong Xing, Feng Kang, Lipeng Zhang, Mingming Qin, Caiyan Liaoning Univ Technol Sch Elect Engn Jinzhou 121001 Peoples R China Harbin Inst Technol Sch Mech Engn & Automat Shenzhen 518055 Peoples R China

When using the Transformer model for wind power prediction, the accuracy of the model predictions tends to be reduced due to the shift in the wind power data distribution, channel mixing, and the inability of the model to establish strong correlations. To address these challenges, this paper proposes an ultra-short-term wind power prediction model based on the DT-DSCTransformer. First, the model applies DT's self-learning standardization and de-standardization parameters to standardize the input and de-standardize the output, mitigating the impact forecasting of data distribution shifts on prediction accuracy. Second, the proposed De-Stationary Channel Attention (DSCAttention) mechanism is introduced. By incorporating De-Stationary Attention (DSAttention) into the channel attention mechanism while maintaining channel independence, the model establishes stronger inter-channel correlations, addressing the performance degradation caused by channel mixing and weak correlations. Finally, experimental analysis demonstrates that the proposed model achieves the highest prediction accuracy compared to commonly used time series forecasting models.

关键词： Predictive models Data models Wind power generation Accuracy Forecasting Transformers Correlation prediction algorithms Computational modeling Wind speed Transformer wind power prediction distribution shift DT DSCAttention

来源：评论

学校读者我要写书评

暂无评论

Enhanced Performance of New Scaling-Free CORDIC for Memory-Based Fast Fourier Transform Architecture

引用

IEEE ACCESS 2025年 13卷 19828-19844页

作者： Paramasivam, C. Chauhan, Sandeep Singh Meena, Veerpratap Sreejagathi, A. Hasini, B. A. V. N. Kishore, K. L. K. Vamsikrishna, T. V. N. G. Sai, M. Durga Ananta Didi, Abdessamad Amrita Vishwa Vidyapeetham Amrita Sch Engn Dept Elect & Commun Engn Bengaluru 560035 India Natl Inst Technol Jamshedpur Dept Elect Engn Jamshedpur 831014 Jharkhand India Natl Energy Ctr Nucl Sci & Technol Rabat 10001 Morocco

Coordinate rotation digital computer (CORDIC) algorithm is an iterative method and it performs the vector rotation operation by micro-rotation with scaling operation in each iteration. This study introduces a high-performance power-efficient new scaling-free coordinate rotation digital computer (NSF-CORDIC) algorithm to perform vector rotation operation in circular coordinates. The scaling operation required in the existing algorithm has been completely removed by the fourth-order approximation of Taylor series (TS). The number of iterations is reduced by an optimized shift value prediction technique for known and fixed angles, like the twiddle factor angle available in the fast Fourier transform (FFT) algorithm. The angle of convergence (AOC) of the algorithm is 57.1 degrees, and it is extended to 180 degrees using the pre-rotation operation and optimized shift value prediction technique. In addition to that, the new CORDIC cell architecture is designed to perform Taylor series-based iterations. Further, a memory-based FFT architecture is designed using a new CORDIC cell. In the FFT architecture, the CORDIC cell performs all the multiplication of twiddle factor. Therefore, the complex constant multiplier required to execute the twiddle factor multiplication is eliminated. The proposed architectures are implemented in the Zynq-7ZC706 FPGA board using the Vivado EDA tool. The scaling-free CORDIC cell architecture has 31% and 3% slice reduction, 72% and 45% slice delay product reduction, and 38% and 23% power reduction compared to the two different existing scaling-free CORDIC designs. The memory-based FFT architecture has 37.8% and 19.6% slice reduction, 38.4% and 27% slice delay product reduction, and 45% and 34% power reduction compared to the two different existing memory-based FFT architecture.

关键词： Computer architecture Fast Fourier transforms Vectors Approximation algorithms Microprocessors Read only memory prediction algorithms Transforms Memory architecture Delays Coordinate rotation digital computer (CORDIC) fast Fourier transform (FFT) memory-based FFT architecture new scaling-free CORIDC micro-rotation scaling operation

来源：评论

学校读者我要写书评

暂无评论

Predicting the Compressive Strength of Recycled Concrete Using Ensemble Learning Model

引用

IEEE ACCESS 2025年 13卷 2958-2969页

作者： Pan, Binghao Liu, Wensheng Zhou, Pan Wu, Dapeng Oliver Hubei Univ Technol Sch Civil Engn Architecture & Environm Wuhan 430068 Hubei Peoples R China Huazhong Univ Sci & Technol Hubei Engn Res Ctr Big Data Secur Sch Cyber Sci & Engn Hubei Key Lab Distributed Syst Secur Wuhan 430074 Hubei Peoples R China City Univ Hong Kong Dept Comp Sci Hong Kong Peoples R China

This research proposes a stacking machine learning method to accurately predict the compressive strength of recycled concrete. The model integrates eXtreme Gradient Boosting (XGBoost), Extra Trees (ET), Decision Tree (DT), and Linear Regression (LR) models, aiming to maximize the prediction accuracy of concrete compressive strength. The model was evaluated using a combination of 63 self-made recycled concrete datasets and 1030 concrete datasets from the UCI Machine Learning Repository. Through optimization based on SHAP values, the new model achieved statistical metrics of RMSE = 1.969, MAE = 1.113, and R-2 = 0.987. The comparison and analysis with the existing work show that this method has excellent performance. Additionally, the feature importance analysis based on SHAP values identified the key input variables affecting concrete compressive strength and improved the model's prediction performance.

关键词： prediction algorithms Stacking Concrete Predictive models Random forests Training Decision trees Boosting Aggregates Accuracy ensemble learning machine learning recycled concrete aggregate strength prediction

来源：评论

学校读者我要写书评

暂无评论

Large Range Curvature Measurement Based on a Cascaded Hollow-Core Fiber Sensor Combining With a PCA plus SVR Demodulation Algorithm

引用

IEEE SENSORS JOURNAL 2025年第9期25卷 15136-15142页

作者： Luo, Junhua Lou, Shuqin Cao, Jiaqi Liu, Ang Guo, Yuying Wang, Zixia Wang, Xin Sheng, Xinzhi Beijing Jiaotong Univ Sch Elect & Informat Engn Key Lab Commun & Informat Syst Beijing Municipal Commiss Educ Beijing 100044 Peoples R China Sch Phys Sci & Engn Beijing 100044 Peoples R China

A novel sensor structure composed of a negative curvature hollow-core fiber and a hollow-core fiber is proposed. The highest sensitivity with respect to wavelength reaches 11.21 nm/m(-1) in the curvature range from 2.37 to 4.49 m(-1), while that with respect to transmission intensity is 11.01 dB/m(-1) in the curvature range from 6.82 to 7.62 m(-1). To expand the curvature measurement range, the support vector regression algorithm is introduced for the prediction of curvature by training and learning the transmission spectra. Combining with the principal component analysis algorithm, which is used for data dimensions reduction, high prediction accuracy can be realized in a large curvature measurement range. The experimental results demonstrate that the mean absolute error and mean squared error for the predicted curvature are as low as 1.75 x 10(-3) m(-1) and 4.98 x 10(-6), respectively, within the curvature range from 0 to 12.78 m(-1). The prediction time is just 0.223 s to predict all 128 curvatures within the measurement range. Moreover, the prediction efficiency could be further improved, as the high prediction accuracy can be maintained even if the sampling rate of the spectra is reduced.

关键词： Optical fiber sensors Sensors Sensitivity Principal component analysis Accuracy Wavelength measurement Polynomials Optical fiber theory Interference prediction algorithms Curvature optical fiber sensor principal component analysis (PCA) support vector regression (SVR)

来源：评论

学校读者我要写书评

暂无评论

Predictive Control of Output-Current in a Single-Pulse-Operated Switched Reluctance Generator

引用

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS 2025年第4期72卷 3414-3424页

作者： Verma, Anupam Narayanan, Gopalaratnam Indian Inst Sci Dept Elect Engn Bangalore 560012 India

This article deals with the output-current control of a single-pulse-operated switched reluctance generator (SRG). Typically, a linear proportional integral (PI) controller whose gain is obtained by hit-and-trial or by a tuning method is utilized. However, due to the highly nonlinear nature of single-pulse operated SRG, the performance of the PI controller tuned at one operating point deteriorates at another. This article proposes a nonlinear predictive control algorithm for output-current control of single-pulse operated SRG. In particular, two independent controllers are developed, one controlling the turn-ON angle and the other controlling the turn-OFF angle. These controllers manifest good steady-state tracking, dynamic response, and disturbance rejection. Compared to a PI controller, the proposed controller performs consistently at different operating points without tuning any controller parameter. The proposed control algorithm does not require prior measurement of machine characteristics. It does not require storing any offline computed data.

关键词： prediction algorithms Predictive control PI control Switches Tuning Transfer functions Optimization Predictive control algorithm single-pulse mode small-signal analysis switched reluctance generator (SRG) turn-OFF angle controller turn-ON angle controller

来源：评论

学校读者我要写书评

暂无评论

Robust Pedestrian Tracking With Severe Occlusions in Public Spaces Using 3D Point Clouds

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2025年第6期26卷 8411-8423页

作者： Ukyo, Riki Amano, Tatsuya Rizk, Hamada Yamaguchi, Hirozumi Moriya, Takao Osaka Univ Dept Informat Sci & Technol Suita 5650871 Japan RIKEN Ctr Computat Sci Kobe 6500047 Japan Tanta Univ Dept Comp & Control Engn Tanta 31733 Egypt HULIX Inc Suita 5650871 Japan

The increasing need for pedestrian tracking within public and commercial domains necessitates innovative solutions that respect privacy concerns. Traditional surveillance methods employing RGB video footage for the monitoring of pedestrian movements have raised significant privacy issues, attributed to their capability to easily identify individuals. This paper shifts focus towards the utilization of 3D point cloud data as a less intrusive alternative for pedestrian monitoring, which inherently minimizes the risk of individual identification. Notwithstanding, traditional tracking methodologies frequently encounter difficulties in environments characterized by high pedestrian density, where occlusions caused by individuals obscuring one another from the sensors' perspective are commonplace. This research introduces an advanced approach leveraging 3D point cloud data to address the challenges inherent in pedestrian tracking. A particular obstacle addressed is the frequent obstruction caused by pedestrians, who may be accompanied by objects such as baby strollers and/or luggage, thus complicating their detection and tracking when positioned behind others or in close proximity to the sensors. Such scenarios significantly undermine the efficacy of pedestrian segmentation and tracking, particularly in the context of Kalman-Filter-based methodologies. To mitigate these challenges, the proposed solution incorporates a method to spatially enhance the missing segments within the framework of Kalman-Filter-based multi-object tracking. The effectiveness of this approach has been rigorously assessed through the application of 3D point cloud data derived from diverse real-world settings, including laboratory environments, shopping mall entrances, and a densely populated urban area in Kita-ku, Osaka City, during a festival event.

关键词： Pedestrians Three-dimensional displays Point cloud compression Sensors Tracking Laser radar Accuracy Urban areas prediction algorithms Object recognition 3D point cloud depth sensor LiDAR Kalman-filter-based multi object tracking occlusion

来源：评论

学校读者我要写书评

暂无评论

TIRDH: A Novel Three-Shadow-Image Reversible Data Hiding Algorithm Using Weight and Modulo

引用

IEEE ACCESS 2025年 13卷 49232-49248页

作者： Chin, Li-Chian Li, Yu-Chen Hsieh, Hung-Mo Wang, Chung-Ming Natl Chung Hsing Univ Dept Comp Sci & Engn Taichung 40223 Taiwan Natl Taiwan Univ Dept Comp Sci & Informat Engn Taipei 106319 Taiwan

Reversible data hiding (RDH) in image media is crucial for securely embedding confidential information while ensuring complete recovery of both the cover image and the hidden data. Traditional RDH techniques often struggle to balance embedding capacity and image fidelity, limiting their practical effectiveness. This paper introduces TIRDH, a novel three-shadow-image RDH scheme that employs weight and modulo operations to achieve high-capacity embedding while preserving image quality. The method embeds an M-ary secret message within a shadow pixel cluster using a 3-tuple weight vector, modulo operations, and a variation table to guide pixel modifications. This guarantees data integrity without pixel underflow or overflow, thereby avoiding the falling-off boundary problem (FOBP) and ensuring reversibility. Additionally, the scheme offers flexible embedding rates, adjustable through a single parameter and weight vector, making it highly adaptable to various user requirements, image types, and application scenarios. Experimental results from three standard image databases demonstrate its effectiveness, achieving embedding rates ranging from 0.04 to 1.6814 bits per pixel while maintaining high image quality, with an average PSNR ranging from 47.65 to 62.65 dB. Furthermore, its prediction mechanism attains 98.2% accuracy, enabling precise performance estimation before actual message embedding. This predictive capability enhances efficiency by allowing users to anticipate the impact on image quality beforehand. Security assessments confirm strong resilience against RS steganalysis and pixel difference histogram attacks, providing enhanced protection for confidential data. Comparisons with 12 state-of-the-art methods show that the proposed approach outperforms competitors when embedding fewer than 50,000 secret bits and surpasses another set of 12 schemes when the embedding rate exceeds 1.0 bits per pixel. For 1,000 test images with an embedding rate of 1.4 bits per pixel,

关键词： Image quality Security prediction algorithms Vectors Standards Media Robustness Histograms Clustering algorithms Indexes Reversible data hiding three-shadow-image weights modulus operator prediction embedding rate embedding capacity image quality

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：