检索结果-内蒙古大学图书馆

Distributed stochastic source seeking and formation control based on delayed measurements

science China(Information sciences) 2025年第3期68卷 319-337页

作者： Pei-Pei ZHANG Lin-Yu YANG Shu-Jun LIU School of Mathematics Sichuan University School of Data and Computer Science Shandong Women's University

In this paper, we design a distributed stochastic source seeking algorithm based on time-delay measurements to implement source seeking and formation control, so that vehicles can achieve and maintain a specific formation during the source seeking process. First, we present continuous-time stochastic averaging theorems for nonlinear delay-differential systems with stochastic perturbations. Then, based on the stochastic extremum seeking method and the leaderless formation strategy,we design a distributed stochastic source seeking algorithm based on time-delay measurements to navigate multiple velocity-actuated vehicles to search for an unknown source while achieving and maintaining a predefined formation, and the effect of the delay is eliminated by adopting the one-stage sequential predictor approach. Moreover, based on our developed stochastic averaging theorems, we prove that the average position of vehicles exponentially converges to a small neighborhood of the source in the almost sure sense, and vehicles can achieve and maintain a predefined formation. Finally, we provide numerical examples to verify the effectiveness of our proposed algorithm.

关键词： distributed source seeking formation control stochastic extremum seeking stochastic averaging time delays

来源：评论

学校读者我要写书评

暂无评论

Improving Multi-model Anomaly Traffic Detection in MEC Networks with Large-Model-powered Continuous Learning

引用

IEEE Network 2025年第3期39卷 56-62页

作者： Zhang, Junjie Chen, Zheyi Cheng, Hongju Li, Jie Min, Geyong Fuzhou University College of Computer and Data Science China Shanghai Jiao Tong University Department of Computer Science and Engineering China University of Exeter Department of Computer Science China

Anomaly traffic detection offers essential technical support for securing Mobile Edge Computing (MEC) networks. The emerging Large Model (LM) has attracted much attention for their excellent data generation and processing capabilities, but it is difficult to deploy LM-based detection models in resource-constrained MEC networks. Existing solutions usually compress large models into tiny ones, but they tend to be impacted by data drift, resulting in decreased detection accuracy. To address this key challenge, we propose CL4Det, a novel multi-model anomaly traffic detection framework with LM-powered continuous learning, where the tiny models deployed in MEC networks can achieve the desired performance comparable to the large models via continuous retraining. Specifically, CL4Det periodically evaluates the model performance degradation caused by data drift in MEC networks and decides whether to generate retraining tasks and their configurations. Meanwhile, CL4Det schedules all traffic detection and retraining tasks with proper resource allocation, aiming to ensure real-time detection and maximize model accuracy. A case study with real-world traffic datasets verifies the effectiveness and superiority of CL4Det. Finally, we outline the challenges and future directions to fully exploit the collaborative potentials of MEC networks and LM in anomaly traffic detection. © 1986-2012 IEEE.

关键词： Resource allocation

来源：评论

学校读者我要写书评

暂无评论

Using Multi-Scale Convolution Fusion and Memory-Augmented Adversarial Autoencoder to Detect Diverse Anomalies in Multivariate Time Series

引用

Tsinghua science and Technology 2025年第1期30卷 234-246页

作者： Zefei Ning Hao Miao Zhuolun Jiang Li Wang College of Computer Science and Technology(College of Data Science) Taiyuan University of TechnologyJinzhong 030600China

Time series anomaly detection is an important task in many applications,and deep learning based time series anomaly detection has made great ***,due to complex device interactions,time series exhibit diverse abnormal signal shapes,subtle anomalies,and imbalanced abnormal instances,which make anomaly detection in time series still a *** and analysis of multivariate time series can help uncover their intrinsic spatio-temporal characteristics,and contribute to the discovery of complex and subtle *** this paper,we propose a novel approach named Multi-scale Convolution Fusion and Memory-augmented Adversarial AutoEncoder(MCFMAAE)for multivariate time series anomaly *** is an encoder-decoder-based framework with four main ***-scale convolution fusion module fuses multi-sensor signals and captures various scales of temporal ***-attention-based encoder adopts the multi-head attention mechanism for sequence modeling to capture global context *** module is introduced to explore the internal structure of normal samples,capturing it into the latent space,and thus remembering the typical ***,the decoder is used to reconstruct the signals,and then a process is coming to calculate the anomaly ***,an additional discriminator is added to the model,which enhances the representation ability of autoencoder and avoids *** on public datasets demonstrate that MCFMAAE improves the performance compared to other state-of-the-art methods,which provides an effective solution for multivariate time series anomaly detection.

关键词： multivariate time series anomaly detection AutoEncoder(AE) multi-scale fusion

来源：评论

学校读者我要写书评

暂无评论

On hierarchical clustering-based identification of PWA model with model structure selection and application to automotive actuators for HiL simulation

引用

Soft Computing 2025年第2期29卷 543-558页

作者： Ren, Zhenxing College of Computer Science and Technology and College of Data Science Taiyuan University of Technology Shanxi Jinzhong China

The application of the electronic control unit (ECU) motivates dynamic models with high precision to simulate mechatronic systems for various analysis and design tasks like hardware-in-the-loop (HiL) simulation. Unlike traditional physical models which are established based on the research experience or the mechanics mechanism, in this study, a novel data-driven modeling approach is presented based on the piecewise affine (PWA) identification method. In this work, the highly nonlinear dynamic of automotive actuators is well approximated by the PWA model. To obtain experimental data that can accurately reflect the characteristics of actuators, a test bench was first built. On this basis, the PWA identification of automotive is composed of the data clustering, the model structure selection, and the model parameter estimation. The proposed clustering method improves the widely used balanced iterative reducing and clustering using hierarchies (BIRCH) by introducing a refinement phase for handling clusters with arbitrary shapes. The model structure selection and the parameter estimation are jointly solved by using the optimization method. The presented method is demonstrated with an academic example and an automotive throttle, and the results show that the proposed method can achieve a high model quality, which means that the normalized root mean squared error (NRMSE) is 0.03 and the absolute maximal prediction error can reach 2.43°. Compared to other models like a physical model or a fuzzy model, the improvement using the proposed model can be up to 50%. Thus, the quality of the proposed PWA model is sufficient for HiL simulation. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2025.

关键词： Hardware-in-the-loop simulation

来源：评论

学校读者我要写书评

暂无评论

A survey on cross-user federated recommendation

引用

science China(Information sciences) 2025年第4期68卷 7-32页

作者： Enyue YANG Yudi XIONG Wei YUAN Weike PAN Qiang YANG Zhong MING College of Computer Science and Software Engineering Shenzhen University School of Electrical Engineering and Computer Science The University of Queensland WeBank AI Lab WeBank Department of Computer Science and Engineering Hong Kong University of Science and Technology College of Big Data and Internet Shenzhen Technology University Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ)

Recommender systems are effective in mitigating information overload, yet the centralized storage of user data raises significant privacy concerns. Cross-user federated recommendation(CUFR) provides a promising distributed paradigm to address these concerns by enabling privacy-preserving recommendations directly on user devices. In this survey, we review and categorize current progress in CUFR, focusing on four key aspects: privacy, security, accuracy, and efficiency. Firstly,we conduct an in-depth privacy analysis, discuss various cases of privacy leakage, and then review recent methods for privacy protection. Secondly, we analyze security concerns and review recent methods for untargeted and targeted *** untargeted attack methods, we categorize them into data poisoning attack methods and parameter poisoning attack methods. For targeted attack methods, we categorize them into user-based methods and item-based methods. Thirdly,we provide an overview of the federated variants of some representative methods, and then review the recent methods for improving accuracy from two categories: data heterogeneity and high-order information. Fourthly, we review recent methods for improving training efficiency from two categories: client sampling and model compression. Finally, we conclude this survey and explore some potential future research topics in CUFR.

关键词： cross-user federated recommendation federated recommendation federated learning recommender systems user privacy

来源：评论

学校读者我要写书评

暂无评论

Enhancing Entity Relationship Extraction in Dialogue Texts Using Hypergraph and Heterogeneous Graph

引用

Chinese Journal of Electronics 2025年第1期34卷 295-308页

作者： Shunmiao Zhang Siyuan Zheng Degen Huang Dan Li School of Computer Science and Mathematics Fujian University of Technology Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology School of Computer Science and Technology Dalian University of Technology Elsevier

Dialogue-based relation extraction(DialogRE) aims to predict relationships between two entities in dialogue. Current approaches to dialogue relationship extraction grapple with long-distance entity relationships in dialogue data as well as complex entity relationships, such as a single entity with multiple types of connections. To address these issues, this paper presents a novel approach for dialogue relationship extraction termed the hypergraphs and heterogeneous graphs model(HG2G). This model introduces a two-tiered structure, comprising dialogue hypergraphs and dialogue heterogeneous graphs, to address the shortcomings of existing methods. The dialogue hypergraph establishes connections between similar nodes using hyper-edges and utilizes hypergraph convolution to capture multi-level features. Simultaneously, the dialogue heterogeneous graph connects nodes and edges of different types, employing heterogeneous graph convolution to aggregate cross-sentence information. Ultimately, the integrated nodes from both graphs capture the semantic nuances inherent in dialogue. Experimental results on the DialogRE dataset demonstrate that the HG2G model outperforms existing state-of-the-art methods.

关键词： Convolution Aggregates Semantics Feature extraction data mining

来源：评论

学校读者我要写书评

暂无评论

Simulation Study: Trajectory Tracking Control Of Parafoil System Based On Deep Reinforcement Learning

Journal of Applied Science and Engineering

引用

Journal of Applied science and Engineering 2025年第12期28卷 2439-2449页

作者： Zhang, Xuepu School of Big Data and Computer Science Guizhou Normal University Guiyang550025 China

The parafoil system is nonlinear and complex with a large time delay. This makes it challenging for traditional control methods to control the parafoil system effectively. However, the Markov property of reinforcement learning offers a new possibility for controlling the parafoil system. Therefore, this paper employs the deep reinforcement learning (DRL) method to train a neural network controller for controlling the parafoil system, based on a modified deterministic version of the distributional soft actor-critic with three refinements (DSAC-T) algorithm and it is named MC-DSAC-T. The controller of the parafoil system is denoted as a multilayer perceptron (MLP) and the objective function of the policy introduces cumulative discounted rewards of a single episode to improve the stability of the iterative update of the policy, a Monte Carlo (MC) thinking. In addition, wind disturbances are introduced during training to enhance the robustness of the neural network controller. First, a nine-degree-of-freedom (nine-DOF) dynamic model of the parafoil system is developed. Secondly, the network structure of the MC-DSAC-T algorithm and the process of updating the network using sampling data were introduced. Finally, the control effects of the neural network controller trained by the proposed method were compared with those of the proportion integration differentiation (PID) controller in a wind environment. While tracking 100 randomly selected trajectory segments, the results show that the neural network controller is superior to the PID controller in distance control accuracy, which proves that the neural network controller can control the parafoil system to perform the tracking task and verify the effectiveness of the proposed method. © The Author(’s).

关键词： Reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Prompt Learning for Few-Shot Question Answering via Self-Context data Augmentation

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on Artificial Intelligence 2025年第3期6卷 589-603页

作者： Qiu, Jian-Qiang Zhang, Chun-Yang Chen, C. L. Philip Fuzhou University College of Computer and Data Science Fuzhou350108 China South China University of Technology School of Computer Science and Engineering Guangzhou510006 China

Pretrained language models (PLMs) have shown remarkable performance on question answering (QA) tasks, but they usually require fine-tuning (FT) that depends on a substantial quantity of QA pairs. Therefore, improving the performance of PLMs in scenarios with only a small number of training examples, also known as a few-shot setting, is of great practical significance. Current mitigation strategies for the few-shot QA task largely rely on pretraining a QA task-specific language model from scratch, overlooking the potential of foundational PLMs to generate QA pairs, particularly in the few-shot setting. To address this issue, we propose a prompt-based QA data augmentation method aimed at automating the creation of high-quality QA pairs. It employs the PFT method, adapting the question generation process of PLMs to the few-shot setting. Additionally, we introduce a dynamic text filling training strategy. This strategy simulates the progressive human learning process, thereby alleviating overfitting of PLMs in the few-shot setting and enhancing their reasoning capability to tackle complex questions. Extensive experiments demonstrate that the proposed method outperforms existing approaches across various few-shot configurations. 2691-4581 © 2024 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See https://***/publications/rights/*** for more information.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Identification of Human Activity from Video Streaming Smartphone data Using Intensified VGG16

引用

International Journal of Engineering, Transactions B: Applications 2025年第6期38卷 1340-1352页

作者： Yadav, R.K. Daniel, A. Semwal, V.B. Department of Computer Science and Engineering Amity University Gwalior India Data Science and Engineering Manipal University Jaipur India Department of Computer Science and Engineering Maulana Azad National Institute Technology Bhopal India

Human activity recognition (HAR) techniques pick out and interpret human behaviors and actions by analyzing data gathered from various sensor devices. HAR aims to recognize and automatically categorize human activities using patterns and attributes taken from sensor data. HAR is complex in implementing the algorithm for a self-recorded dataset, including challenges such as age variation, wearing different clothes, environment and surface, the direction of the smartphone camera, and many more. The paper aims to propose a VGG16 deep learning framework including an activation function and different optimizers for classifying human activity from the real-time captured dataset;further, we compare the evaluated results with existing results. The proposed methods achieved 99.88% accuracy with excellent precision, recall, and F_measure values. Comparing the evaluated result with existing outcomes over the WISDM and UCI-HAR datasets. The new things in the article are a self-captured dataset of various aged male, female, and healthy volunteers to perform seven activities. Furthermore, this research uses Tensor Processing Units (TPU) available on Kaggle to improve classification accuracy while reducing error rates and speeding up execution. ©2025 The author(s).

关键词： Smartphones

来源：评论

学校读者我要写书评

暂无评论

CoverGAN: cover photo generation from text story using layout guided GAN

引用

Soft Computing 2025年第1期29卷 405-423页

作者： Cheema, Adeel Naeem, M. Asif Department of Data Science and Artificial Intelligence National University of Computer and Emerging Sciences Islamabad Pakistan

Generating cover photos from story text is a non trivial challenge to solve. Existing approaches focus on generating only images from given text prompt. To the best of our knowledge, non of these approaches focus on generating cover photos from a text story. The paper addresses this issue by introducing multi-object image generation with text title from a text story. We split the problem into three steps:- understanding semantics of text story, predicting layout of objects, and generating a cover photo. At start, a semantic relation was encoded between text story objects using a Scene graph, then features from graph neural network were concatenated with single object features from scene graph to create an object layout. All of these features were then passed on to the image generating part. Image generation was further divided into two phases. In the first phase, the image is generated using a scene graph image generation model. While in the second phase, the results of first phase were further enhanced using image translation model conditioned on the object layout. In final phase we generated title of the given story based on generated image. In our experiments, we used custom dataset of text stories with three animal categories along with the COCO dataset. For the image generating part, we evaluated our approach with state of the art models known as scene_gen and sg2im. Our method generated high-resolution informative cover photo with story title by positioning the objects at right locations as specified in the text story. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2025.

关键词： Graph neural networks

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：