检索结果-内蒙古大学图书馆

Data Driven Control and Learning Systems (DDCLS)

作者： Xiaoxue Liang Kuangrong Hao Lei Chen Lihong Ren Engineering Research Center of Digitized Textile and Apparel Technology Ministry of Education Donghua University Shanghai China College of Information Sciences and Technology Donghua University Shanghai China

ISBN: (数字)9798350361674

ISBN: (纸本)9798350361681

In the real world, time series data are ubiquitous, and the prediction task of time series data is very important. Most real-world time series data do not satisfy the assumption of independent and identical distribution (i.i.d.) due to environmental changes, that is, the distribution of the training datasets is different from the distribution of the test datasets, $P(X_{i})= P(X_{j})$ , but the conditional distribution is usually considered to be unchanged, $P(y\vert x_{i})= P(y\vert x_{j})$ . In this case, it is defined as covariate drift. However, most of the existing prediction algorithms are based on the assumption of i.i.d., so these algorithms have great limitations in the prediction of time series data under covariant drift. Therefore, the AEIF-MLP model is proposed, a time series data prediction model based on MLP and causal structures. The model mainly consists of two modules. Based on the principle of maximum entropy, we propose an adaptive environment segmentation module to separate different environments in the training datasets. Based on the causal structure, we propose an invariant feature learning module to learn common invariant features in different environments to train the model to deal with test datasets of unknown distribution. In summary, this method solves the problem of time series prediction under covariate drift. The validity of the model is verified by drift data sets in different environments for the first time, and the validity of the model is further verified on two real data sets.

关键词： Training Representation learning Learning systems Adaptation models Time series analysis Predictive models Prediction algorithms

来源：评论

学校读者我要写书评

暂无评论

A Dense Feature Iterative Fusion Network for Extracting Building Contours from Remote Sensing Imagery

引用

Journal of Donghua University(English Edition) 2024年第6期41卷 654-661页

作者： WU Jiangyan WANG Tong College of Information Science and Technology Donghua UniversityShanghai 201620China Engineering Research Center of Digitized Textile&Apparel Technology Ministry of EducationDonghua UniversityShanghai 201620China

Extracting building contours from aerial images is a fundamental task in remote *** building extraction methods cannot accurately extract building contour information and have errors in extracting small-scale *** paper introduces a novel dense feature iterative(DFI)fusion network,denoted as DFINet,for extracting building *** network uses a DFI decoder to fuse semantic information at different scales and learns the building contour knowledge,producing the last features through iterative *** dense feature fusion(DFF)module combines features at multiple *** employ the contour reconstruction(CR)module to access the final *** experiments validate the effectiveness of the DFINet on two different remote sensing datasets,INRIA aerial image dataset and Wuhan University(WHU)building *** the INRIA aerial image dataset,our method achieves the highest intersection over union(IoU),overall accuracy(OA)and F 1 scores compared to other state-of-the-art methods.

关键词： remote sensing image building contour extraction feature iteration

来源：评论

学校读者我要写书评

暂无评论

M2S2: A Policy Transfer Framework for Deep Reinforcement Learning

M2S2: A Policy Transfer Framework for Deep Reinforcement Lea...

引用

Chinese Automation Congress (CAC)

作者： Huashan Liu Dezhen Yin Xiangjian Li College of Information Science and Technology Donghua University Shanghai Engineering Research Center of Digitized Textile & Apparel Technology Ministry of Education Shanghai

Deep reinforcement learning (DRL) has greatly promoted the progress of investigations in robotic motion planning. However, when faced with different scenarios, the path obtained from each training for a specific scenario is unique, which means that the DRL needs additional resources to relearn and adapt the policy to the new scenario. Inspired by the mentor-student relationship, a multi-mentor-single-student (M2S2) framework is proposed in this paper to enhance the policy generalization performance among similar task scenarios. It promotes the learning efficiency of the student network by adding networks that have been trained in similar scenarios as the mentor networks in the training, where weight parameters are assigned to balance the relationship among all the networks. The experimental results show that M2S2 has good feasibility and superior performance for policy transfer of DRL among similar scenarios.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prescribed Time Cluster Formation for Multi-Agent Systems under Cooperation-Competition Interaction 42

Prescribed Time Cluster Formation for Multi-Agent Systems un...

引用

42nd Chinese Control Conference, CCC 2023

作者： Li, Hongliang Zhou, Wuneng Hu, Fei Liao, Kaili College of Information Science and Technology Donghua University Shanghai201620 China Engineering Research Center of Digitized Textile and Fashion Technology Ministry of Education Donghua University Shanghai201620 China

ISBN: (纸本)9789887581543

In this paper, the problem of prescribed time cluster formation of multi-agent systems with cooperation-competition interaction is studied. Firstly, the concept of prescribed time cluster formation of multi-agent systems with cooperation-competition interaction is introduced. Secondly, an state feedback controller is designed to ensure the prescribed time cluster formation by Lyapunov functional method and free-weight matrix technology. Finally, a numerical example is given to illustrate the effective of the the results and method of this paper. © 2023 Technical Committee on Control Theory, Chinese Association of Automation.

关键词： Multi agent systems

来源：评论

学校读者我要写书评

暂无评论

A multi-scale dual moving horizon estimation method for state of charge of lithium-ion battery

A multi-scale dual moving horizon estimation method for stat...

引用

Chinese Control Conference (CCC)

作者： Jiacheng Wen Xin Cai Engineering Research Center of Digitized Textile & Apparel Technology Ministry of Education College of Information Science and Technology Donghua University Shanghai China

For the battery management system (BMS), accurate state of charge (SOC) estimation plays an important role in enhancing efficiency and ensuring safety of battery. Considering the tradeoff between accuracy of SOC and computational complexity, this paper proposes a multi-scale dual moving horizon estimation (MHE) algorithm, which incorporates a SOC estimator and a model parameters estimator using different time scales. Firstly, the state-space models of the SOC and parameters are built on basis of the equivalent circuit model and parametric sensitivity analysis, respectively. Secondly, the estimation problem of the model parameters is transformed into an optimization problem solved in a larger time scale used for the update of the SOC. Thirdly, the estimation problem of the SOC is solved in a smaller time scale used for the update of the model parameters. Finally, simulation results show that this method can effectively enhance accuracy and reduce computational burden compared with the methods such as MHE and dual EKF. The effect of the MHE window size on the SOC estimation is also discussed.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Mitigating Poisoning Attacks in Federated Learning Through Deep One-Class Classification

引用

IEEE Transactions on Cognitive Communications and Networking 2025年

作者： Zhang, Anqi Zhao, Ping Lu, Wenke Zhou, Yipeng Zhang, Wenqian Zhang, Guanglin Donghua University College of Information Science and Technology Shanghai201620 China Ministry of Education Engineering Research Center of Digitized Textile and Apparel Technology Shanghai201620 China Macquarie University School of Computing Faculty of Science and Engineering 2122 Australia

Federated learning (FL) enables clients to learn a machine learning model collaboratively without sharing their private local data to the server. However, due to its distributed structure, FL is vulnerable to poisoning attacks where adversaries intentionally send the poisoned local model parameters to the server and further affect the behavior of the global model. Existing works on mitigating poisoning attacks in FL are difficult to accurately characterize data relationships of high-dimensional parameters. Moreover, they cannot guarantee Byzantine robustness when the majority of clients are compromised (Byzantine ratio) and clients’ local datasets are highly non-independent and identically distributed (non-IID). In this paper, we conduct a pioneering work to introduce deep one-class classification into mitigating poisoning attacks in FL, which can guarantee Byzantine robustness even under a Byzantine ratio and non-IID degree greater than 0.5 with the assistance of a root dataset. Our key idea is to adequately learn vector features of benign local model parameters using Deep Support Vector Data Description (Deep SVDD) and achieve the optimal classification by training a deep learning-based one-class classifier equipped with a proper decision boundary based on the root dataset. To further optimize the classifier, we employ a regularizer based on random noise injections, which can address the hypersphere collapse problem inherent in Deep SVDD. Exhaustive experiments on MNIST, F-MNIST, and CIFAR-10 demonstrate that compared with five typical Byzantine-robust methods, our defense strategy achieves excellent effectiveness in mitigating targeted/untargeted poisoning attacks and an adaptive attack in FL. Even with a similar but different root dataset, it still maintains a good Byzantine robustness. © 2015 IEEE.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

research on Digital Identification of Garment Pieces Based on Contour Fitting Algorithm 18

Research on Digital Identification of Garment Pieces Based o...

引用

18th International Conference on Intelligent Systems and Knowledge engineering, ISKE 2023

作者： Xue, Jiahe Chi, Cheng Wang, Zheng Yang, Shichen Jiang, Xuewei Wuhan Textile and Apparel Digital Engineering Technology Research Center School of Fashion Wuhan Textile University Wuhan China School of Fashion Wuhan Textile University Wuhan China School of Computer Science and Artificial Intelligence Wuhan Textile University Wuhan China

ISBN: (纸本)9798350318401

To help designers analyze the pattern shape of ethnic clothing and obtain its size parameters more conveniently, the contour fitting algorithm was proposed. This algorithm realizes the digital recognition and simulation of complex garment pieces, which combines image edge detection, Fourier descriptor reshaping, garment piece corner point detection, Bezier curve fitting and gradient descent. Compared with the traditional method, the garment piece corner point detection can effectively circumvent the noise generated by the threads of the garment piece. The gradient descent method improved the accuracy of contour fitting by adjusting the control points of Bezier curve. The accuracy of this method was compared the manual method for garment piece size measurement of Mongolian garments. The results show that the errors measured by the algorithm ranged from 0.06 to 0.18 cm, and their average error (0.12 xm) is close to the manual measured error (0.13 cm), This algorithm is in accordance with the GB/T 2660-2017 and can be applied to ethnic clothing restoration design. © 2023 IEEE.

关键词： Errors

来源：评论

学校读者我要写书评

暂无评论

PCB Defect Detection Algorithm Based on Multi-scale Fusion Network

PCB Defect Detection Algorithm Based on Multi-scale Fusion N...

引用

Chinese Intelligent Automation Conference, CIAC 2023

作者： Liao, Xiaofei Su, Xuance Li, Guangyu Chao, Bohang College of Information Science and Technology Donghua University Shanghai201620 China College of Information Science and Technology Engineering Research Center of Digitized Textile and Fashion Technology Ministry of Education Donghua University Shanghai201620 China

ISBN: (纸本)9789819961863

PCB defect detection is a crucial link for the production of PCB board, in order to pursue the yield, the production chain must ensure that there is an efficient PCB defect detection method. The traditional PCB defect detection adopts image processing algorithm, compares the detected image with the standard image, and determines the PCB defect type according to the parameters set. The disadvantage of this method is poor generalization and robustness of detection algorithm. In order to solve the above problems, this paper proposes a PCB defect detection model based on improved CSPDarknet53. Firstly, the PCB dataset is preprocessed, and then the multi-scale feature fusion method is used to complement the information of adjacent feature layers of the network, so as to retain more features. The experimental results show that the average accuracy of the proposed model in PCB defect detection can reach 97.4%, and the average speed of image detection is 43.8ms. The model can effectively identify various types of defects of PCB boards, and has practical application value for the quality detection of industrial PCB boards. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2023.

关键词： Defects

来源：评论

学校读者我要写书评

暂无评论

Adaptive Line-Of-Sight guidance law based on vector fields path following for underactuated unmanned surface vehicle

arXiv

引用

arXiv 2024年

作者： Qi, Jie Wang, Ronghua Wu, Nailong Fan, Yuxin Wang, Jigang College of Information Science and Technology Donghua University Shanghai201620 China Engineering Research Center of Digitized Textile & Apparel Technology Ministry of Education Donghua University Shanghai201620 China

The focus of this paper is to develop a methodology that enables an unmanned surface vehicle (USV) to efficiently track a planned path. The introduction of a vector field-based adaptive line-of-sight guidance law (VFALOS) for accurate trajectory tracking and minimizing the overshoot response time during USV tracking of curved paths improves the overall line-of-sight (LOS) guidance method. These improvements contribute to faster convergence to the desired path, reduce oscillations, and can mitigate the effects of persistent external disturbances. It is shown that the proposed guidance law exhibits Κ-exponential stability when converging to the desired path consisting of straight and curved lines. The results in the paper show that the proposed method effectively improves the accuracy of the USV tracking the desired path while ensuring the safety of the USV work. Copyright © 2024, The Authors. All rights reserved.

关键词： Unmanned surface vehicles

来源：评论

学校读者我要写书评

暂无评论

An Adaptive MPC Trajectory Tracking Algorithm for Autonomous Vehicles 17

An Adaptive MPC Trajectory Tracking Algorithm for Autonomous...

引用

17th International Conference on Computational Intelligence and Security, CIS 2021

作者： Chang, Ge Suqin, Qian College of Information Science and Technology Engineering Research Center of Digitized Textile Apparel Technology Donghua University Ministry of Education Shanghai China

ISBN: (纸本)9781665494892

An adaptive trajectory tracking control algorithm based on the vehicle dynamics model and model predictive control (MPC) is designed to improve the accuracy of trajectory tracking and the ride comfort and vehicle stability of the driver. The algorithm is mainly due to the different changes of transverse speed and longitudinal speed under different road conditions. The weight coefficient matrix in model predictive control is optimized in real-time and online through fuzzy control so that the controller can quickly reduce the tracking error and ensure the accuracy of track tracking when the tracking track changes greatly. When the tracking track changes little, it can reduce the change of the front wheel angle of the vehicle, ensure the stability of the vehicle and improve the comfort of passengers. The results show that the controller can update the weight coefficient in real-time with the change of vehicle speed, and improve the accuracy and stability of vehicle trajectory tracking. © 2021 IEEE.

关键词： Model predictive control

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：