检索结果-内蒙古大学图书馆

HEART: Achieving Timely Multi-Model Training for Vehicle-Edge-Cloud-Integrated Hierarchical Federated Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Yang, Xiaohong Liwang, Minghui Wang, Xianbin Cheng, Zhipeng Hosseinalipour, Seyyedali Dai, Huaiyu Jiao, Zhenzhen School of Informatics Xiamen University Fujian China Department of Control Science and Engineering The National Key Laboratory of Autonomous Intelligent Unmanned Systems Japan Frontiers Science Center for Intelligent Autonomous Systems Ministry of Education Tongji University Shanghai China Department of Electrical and Computer Engineering Western University ON Canada School of Future Science and Engineering Soochow University Suzhou China Department of Electrical Engineering University at Buffalo-SUNY BuffaloNY United States Department of Electrical and Computer Engineering NC State University RaleighNC United States iF-Labs Beijing Teleinfo Technology Company Ltd. CAICT Beijing China

The rapid growth of AI-enabled Internet of Vehicles (IoV) calls for efficient machine learning (ML) solutions that can handle high vehicular mobility and decentralized data. This has motivated the emergence of Hierarchical Federated Learning over vehicle-edge-cloud architectures (VEC-HFL). Nevertheless, one aspect which is underexplored in the literature on VEC-HFL is that vehicles often need to execute multiple ML tasks simultaneously, where this multi-model training environment introduces crucial challenges. First, improper aggregation rules can lead to model obsolescence and prolonged training times. Second, vehicular mobility may result in inefficient data utilization by preventing the vehicles from returning their models to the network edge. Third, achieving a balanced resource allocation across diverse tasks becomes of paramount importance as it majorly affects the effectiveness of collaborative training. We take one of the first steps towards addressing these challenges via proposing a framework for multi-model training in dynamic VEC-HFL with the goal of minimizing global training latency while ensuring balanced training across various tasks—a problem that turns out to be NP-hard. To facilitate timely model training, we introduce a hybrid synchronous-asynchronous aggregation rule. Building on this, we present a novel method called Hybrid Evolutionary And gReedy allocaTion (HEART). The framework operates in two stages: first, it achieves balanced task scheduling through a hybrid heuristic approach that combines improved Particle Swarm Optimization (PSO) and Genetic Algorithms (GA);second, it employs a low-complexity greedy algorithm to determine the training priority of assigned tasks on vehicles. Experiments on real-world datasets demonstrate the superiority of HEART over existing methods. © 2025, CC BY-NC-ND.

关键词： Federated learning

Risk-Averse Certification of Bayesian Neural Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhang, Xiyue Wang, Zifan Gao, Yulong Romao, Licio Abate, Alessandro Kwiatkowska, Marta Department of Computer Science University of Oxford United Kingdom Division of Decision and Control Systems KTH Royal Institute of Technology Sweden Department of Electrical and Electronic Engineering Imperial College London United Kingdom Department of Wind and Energy Systems Technical University of Denmark Denmark

In light of the inherently complex and dynamic nature of real-world environments, incorporating risk measures is crucial for the robustness evaluation of deep learning models. In this work, we propose a Risk-Averse Certification framework for Bayesian neural networks called RAC-BNN. Our method leverages sampling and optimisation to compute a sound approximation of the output set of a BNN, represented using a set of template polytopes. To enhance robustness evaluation, we integrate a coherent distortion risk measure—Conditional Value at Risk (CVaR)—into the certification framework, providing probabilistic guarantees based on empirical distributions obtained through sampling. We validate RAC-BNN on a range of regression and classification benchmarks and compare its performance with a state-of-the-art method. The results show that RAC-BNN effectively quantifies robustness under worst-performing risky scenarios, and achieves tighter certified bounds and higher efficiency in complex tasks. Copyright © 2024, The Authors. All rights reserved.

关键词： Risk assessment

Phase chimera states on nonlocal hyperrings

学校读者我要写书评

暂无评论

Physical Review E 2024年第2期109卷 L022201-L022201页

作者： Riccardo Muolo Thierry Njougouo Lucia Valentina Gambuzza Timoteo Carletti Mattia Frasca Department of Systems and Control Engineering Tokyo Institute of Technology Tokyo 152-8552 Japan Department of Mathematics University of Namur B5000 Namur Belgium naXys Namur Institute for Complex Systems University of Namur B5000 Namur Belgium Faculty of Computer Science University of Namur B5000 Namur Belgium Department of Electrical and Electronic Engineering University of Buea P.O. Box 63 Buea Cameroon Department of Electrical Electronics and Computer Science Engineering University of Catania 95125 Catania Italy Istituto di Analisi dei Sistemi ed Informatica “A. Ruberti” IASI-CNR 00185 Roma Italy

Chimera states are dynamical states where regions of synchronous trajectories coexist with incoherent ones. A significant amount of research has been devoted to studying chimera states in systems of identical oscillators, nonlocally coupled through pairwise interactions. Nevertheless, there is increasing evidence, also supported by available data, that complex systems are composed of multiple units experiencing many-body interactions that can be modeled by using higher-order structures beyond the paradigm of classic pairwise networks. In this work we investigate whether phase chimera states appear in this framework, by focusing on a topology solely involving many-body, nonlocal, and nonregular interactions, hereby named nonlocal d-hyperring, (d+1) being the order of the interactions. We present the theory by using the paradigmatic Stuart-Landau oscillators as node dynamics, and we show that phase chimera states emerge in a variety of structures and with different coupling functions. For comparison, we show that, when higher-order interactions are “flattened” to pairwise ones, the chimera behavior is weaker and more elusive.

关键词： Chimera states Collective behavior in networks Dynamics of networks Pattern formation Synchronization Synchronization transition

Model Reference Adaptive Stabilizing control for Leader-following Consensus

学校读者我要写书评

暂无评论

IEEE Transactions on Automatic control 2025年

作者： Yue, Dongdong Shi, Jiantao Shi, Ling Frasca, Paolo Baldi, Simone Nanjing Tech University College of Electrical Engineering and Control Science Nanjing211816 China Hong Kong University of Science and Technology Department of Electronic and Computer Engineering Clear Water Bay China Univ. Grenoble Alpes Cnrs Inria Grenoble Inp GIPSALab Grenoble38000 France Southeast University School of Mathematics Nanjing210096 China

Complex networks, neuroscience, and other applications have shown examples of multi-agent adaptive systems that must follow (over possibly short times) reference dynamics that are neither Hurwitz nor neutrally stable. However, such leader-following behavior would be impossible with existing adaptive consensus methods, e.g., based on model reference adaptive control (MRAC), since the stability of the reference dynamics is required. To fill this gap, we propose a novel model reference adaptive stabilizing control (MRASC) framework for leader-following consensus of multi-agent systems with unknown and heterogeneous dynamics. Differently from several approaches in the leader-following consensus literature, the proposed framework is free of any extra distributed observer layer for the leader's signal, as the reconstruction of such signals is intrinsic in the adaptive laws. Besides, the framework does not require Hurwitz or neutral stability of the leader and generalizes existing acyclic requirements on the communication graph among the follower. Starting from any weakly connected communication digraph, the proposed method allows to derive a lower bound, useful from the network design point of view, for the minimum number of followers that should be pinned by the leader. © 2025 IEEE.

关键词： Multi agent systems

Generative Artificial Intelligence Assisted Wireless Sensing: Human Flow Detection in Practical Communication Environments

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Jiacheng Du, Hongyang Niyato, Dusit Xiong, Zehui Kang, Jiawen Ai, Bo Han, Zhu Kim, Dong In School of Computer Science and Engineering Nanyang Technological University Singapore639798 Singapore Pillar of Information Systems Technology and Design Singapore University of Technology and Design Singapore China Ministry of Education Guangzhou510006 China State Key Laboratory of Rail Traffic Control and Safety Beijing Engineering Research Center of High-Speed Railway Broadband Mobile Communications Beijing Jiaotong University Beijing100044 China Department of Electrical and Computer Engineering University of Houston HoustonTX77004 United States Department of Electrical and Computer Engineering Sungkyunkwan University Suwon16419 Korea Republic of

Groundbreaking applications such as ChatGPT have heightened research interest in generative artificial intelligence (GAI). Essentially, GAI excels not only in content generation but also signal processing, offering support for wireless sensing. Hence, we introduce a novel GAI-assisted human flow detection system (G-HFD). Rigorously, G-HFD first uses the channel state information (CSI) to estimate the velocity and acceleration of propagation path length change of the human induced reflection (HIR). Then, given the strong inference ability of the diffusion model, we propose a unified weighted conditional diffusion model (UW-CDM) to denoise the estimation results, enabling detection of the number of targets. Next, we use the CSI obtained by a uniform linear array with wavelength spacing to estimate the HIR’s time of flight and direction of arrival (DoA). In this process, UW-CDM solves the problem of ambiguous DoA spectrum, ensuring accurate DoA estimation. Finally, through clustering, G-HFD determines the number of subflows and the number of targets in each subflow, i.e., the subflow size. The evaluation based on practical downlink communication signals shows G-HFD’s accuracy of subflow size detection can reach 91%. This validates its effectiveness and underscores the significant potential of GAI in the context of wireless sensing. © 2024, CC BY.

关键词： Artificial intelligence

State Derivative Normalization for Continuous-Time Deep Neural Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Weigand, Jonas Beintema, Gerben I. Ulmen, Jonas Görges, Daniel Tóth, Roland Schoukens, Maarten Ruskowski, Martin Chair of Machine Tools and Control Systems RPTU Kaiserslautern The German Research Center for Artificial Intelligence Kaiserslautern Germany Group The Department of Electrical Engineering Eindhoven University of Technology Netherlands The Systems and Control Laboratory The Institute for Computer Science and Control Budapest Hungary Institute for Electromobility RPTU Kaiserslautern Germany

The importance of proper data normalization for deep neural networks is well known. However, in continuous-time state-space model estimation, it has been observed that improper normalization of either the hidden state or hidden state derivative of the model estimate, or even of the time interval can lead to numerical and optimization challenges with deep learning based methods. This results in a reduced model quality. In this contribution, we show that these three normalization tasks are inherently coupled. Due to the existence of this coupling, we propose a solution to all three normalization challenges by introducing a normalization constant at the state derivative level. We show that the appropriate choice of the normalization constant is related to the dynamics of the to-be-identified system and we derive multiple methods of obtaining an effective normalization constant. We compare and discuss all the normalization strategies on a benchmark problem based on experimental data from a cascaded tanks system and compare our results with other methods of the identification literature. © 2024, CC BY-SA.

关键词： Ordinary differential equations

Pattern Recognition Method for Detecting engineering Errors on Technical Drawings

学校读者我要写书评

暂无评论

Pattern Recognition Method for Detecting Engineering Errors ...

AI IoT Congress (AIIoT), World

作者： Rimma Dzhusupova Richa Banotra Jan Bosch Helena Holmströ m Olsson Department of Electrical &#x0026 Instrumentation Control &#x0026 Safety Systems McDermott The Hague The Netherlands Department of Instrumentation Control &#x0026 Department of Computer Science and Engineering Chalmers University of Technology Gothenburg Sweden Department of Computer Science and Media Technology Malm&#x00F6 University Malm&#x00F6 Sweden

Many organizations are looking for how to automate repetitive tasks to reduce manual work and free up resources for innovation. Machine Learning, especially Deep Learning, increases the chance of achieving this goal while working with technical documentation. Highly costly engineering hours can be saved, for example, by empowering the manual check with AI, which helps to reduce the total time for technical documents review. This paper proposes a way to substantially reduce the hours spent by process engineers reviewing P&IDs (Piping & Instrumentation Diagrams). The developed solution is based on a deep learning model for analyzing complex real-life engineering diagrams to find design errors - patterns that are combinations of high-level objects. Through the research on an extensive collection of P&ID files provided by McDermott, we prove that our model recognizes patterns representing engineering mistakes with high accuracy. We also describe our experience dealing with class-imbalance problems, labelling, and model architecture selection. The developed model is domain agnostic and can be re-trained on various schematic diagrams within engineering fields and, as well, could be used as an idea for other researchers to see whether similar solutions could be built for different industries.

关键词：

Worst-Case Integrity Attacks and Resilient State Estimation With Partially Secured Measurements

学校读者我要写书评

暂无评论

IEEE Transactions on control of Network Systems 2025年

作者： Zhou, Jing Shang, Jun Chen, Tongwen University of Alberta Department of Electrical and Computer Engineering EdmontonABT6G 1H9 Canada Tongji University Department of Control Science and Engineering Shanghai Institute of Intelligent Science and Technology National Key Laboratory of Autonomous Intelligent Unmanned Systems Frontiers Science Center for Intelligent Autonomous Systems Shanghai200092 China

This paper examines the problem of optimal deception attacks against state estimation with partially secured measurements, where smart sensors transmit innovation sequences to the remote end for information fusion. Due to resource limitations or defensive countermeasures, the adversary can only modify data packets transmitted through unreliable channels. Meanwhile, the attack should be synthesized with sophistication to deceive an anomaly detector. To investigate the vulnerabilities of such estimation systems without feedback links and enhance security performance, the optimal attack policy is derived by formulating and explicitly solving a convex optimization problem, with the goal of maximizing the sum of estimation errors. Subsequently, a novel attack detection and resilient state estimation algorithm is proposed to ensure an acceptable level of estimation accuracy. The theoretical performance metrics, including false alarm rates for the proposed detector, are provided. Finally, the effectiveness of the results is confirmed through numerical examples. © 2014 IEEE.

关键词： Optimal detection

Inverse Kalman Filtering for Systems with Correlated Noises

学校读者我要写书评

暂无评论

Inverse Kalman Filtering for Systems with Correlated Noises

IEEE Conference on Decision and control

作者： Yibei Li Xiaoming Hu Bo Wahlberg Lihua Xie School of Electrical and Electronic Engineering Nanyang Technological University Singapore Department of Mathematics Division of Optimization and Systems Theory KTH Royal Institute of Technology Sweden Division of Decision and Control Systems School of Electrical Engineering and Computer Sciencem KTH Royal Institute of Technology Sweden

This paper focuses on two inverse problems of the Kalman filter in which the process and measurement noises are correlated. The unknown covariance matrix in a stochastic system is reconstructed from observations of its posterior beliefs. For the standard inverse Kalman filtering problem, a novel duality-based formulation is proposed, where a well-defined inverse optimal control (IOC) problem is solved instead. Identifiability of the underlying model is proved, and a least squares estimator is designed that is statistically consistent. The time-invariant case using the steady-state Kalman gain is further studied. Since this inverse problem is ill-posed, a canonical class of covariance matrices is constructed, which can be uniquely identified from the dataset with asymptotic convergence. Finally, the performances of the proposed methods are illustrated by numerical examples.

关键词：