检索结果-内蒙古大学图书馆

The Journal of Machine Learning Research 2024年第1期25卷 1132-1187页

作者： Ali Ramezani-Kebrya Kimon Antonakopoulos Volkan Cevher Ashish Khisti Ben Liang Department of Informatics University of Oslo and Visual Intelligence Centre Integreat Norwegian Centre for Knowledge-driven Machine Learning Oslo Norway Laboratory for Information and Inference Systems (LIONS) EPFL EPFL STI IEL LIONS Lausanne Switzerland Department of Electrical and Computer Engineering University of Toronto Toronto ON Canada

While momentum-based accelerated variants of stochastic gradient descent (SGD) are widely used when training machine learning models, there is little theoretical understanding on the generalization error of such methods. In this work, we first show that there exists a convex loss function for which the stability gap for multiple epochs of SGD with standard heavy-ball momentum (SGDM) becomes unbounded. Then, for smooth Lipschitz loss functions, we analyze a profiles momentum-based update rule, i.e., SGD with early momentum (SGDEM) under a broad range of step-sizes, and show that it can train machine learning models for multiple epochs with a guarantee for generalization. Finally, for the special case of strongly convex loss functions, we find a range of momentum such that multiple epochs of standard SGDM, as a special form of SGDEM, also generalizes. Extending our results on generalization, we also develop an upper bound on the expected true risk, in terms of the number of training steps, sample size, and momentum. Our experimental evaluations verify the consistency between the numerical results and our theoretical bounds. SGDEM improves the generalization error of SGDM when training ResNet-18 on ImageNet in practical distributed settings.

关键词： uniform stability generalization error heavy-ball momentum stochastic gradient descent non-convex

来源：评论

学校读者我要写书评

暂无评论

Joint Optimization for Cooperative Service-Caching, Computation-Offloading, and Resource-Allocations Over EH/MEC-Based Ultra-Dense Mobile Networks

Joint Optimization for Cooperative Service-Caching, Computat...

引用

IEEE International Conference on Communications (ICC)

作者： Zhian Chen Fei Wang Xi Zhang College of Electronic and Information Engineering Southwest University Chongqing China Department of Electrical and Computer Engineering Networking and Information Systems Laboratory Texas A&M University College Station TX USA

Mobile edge-computing (MEC) enabled ultra-dense networks (UDNs), which merges edge-computing with UDNs, can provide enormous benefits, e.g., ultra-low latency. However, due to the ultra-dense deployment of small-cell base stations (SBSs), it becomes infeasible to just depend on the grid power for energy providing, and also it is challenging to jointly optimize service-caching, computation-offloading, and resource-allocation. Integrating energy-harvesting (EH) techniques into MEC-enabled UDNs, we investigate the joint optimization for cooperative service-caching, computation-offloading, and resource-allocation. In our considered UDNs, there exist a large number of EH-based mobile users (MUs) and a mixture of on-grid SBSs, powered by electric grid, and off-grid SBSs, powered by solar, radio frequency (RF) energy, etc. We formulate an energy minimization problem to minimize the sum of weighted energy consumption of all MUs and off-grid SBSs. Also, we develop a two-timescale based joint cooperative service-caching, computation-offloading, and resource-allocation scheme based on the hierarchical multiagent deep reinforcement learning (HMDRL). Using HMDRL, we first derive SBSs' cooperative service-caching policies which are updated in each time frame consisting of multiple time slots. Then, we derive MUs' and SBSs' computation-offloading policies and SBSs' computation resource-allocation policies, which are updated in each time slot. Finally, we validate and evaluate the performances of our proposed schemes through simulations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Age-of-Information and Energy Optimization in Digital Twin Edge Networks

Age-of-Information and Energy Optimization in Digital Twin E...

引用

IEEE Conference on Global Communications (GLOBECOM)

作者： Yongna Guo Yaru Fu Yan Zhang Tony Q. S. Quek School of Electrical Engineering and Computer Science KTH Royal Institute of Technology Stockholm Sweden School of Science and Technology Hong Kong Metropolitan University China Department of Informatics University of Oslo Oslo Norway Information Systems Technology and Design Singapore University of Technology and Design Singapore

ISBN: (数字)9798350351255

ISBN: (纸本)9798350351262

In this paper, we study the intricate realm of digital twin synchronization and deployment in multi-access edge computing (MEC) networks, with the aim of optimizing and balancing the two performance metrics Age of Information (AoI) and energy efficiency. We jointly consider the problems of edge association, power allocation, and digital twin deployment. However, the inherent randomness of the problem presents a significant challenge in identifying an optimal solution. To address this, we first analyze the feasibility conditions of the optimization problem. We then examine a specific scenario involving a static channel and propose a cyclic scheduling scheme. This enables us to derive the sum AoI in closed form. As a result, the joint optimization problem of edge association and power control is solved optimally by finding a minimum weight perfect matching. Moreover, we examine the one-shot optimization problem in the contexts of both frequent digital twin migrations and fixed digital twin deployments, and propose an efficient online algorithm to address the general optimization problem. This algorithm effectively reduces system costs by balancing frequent migrations and fixed deployments. Numerical results demonstrate the effectiveness of our proposed scheme in terms of low cost and high efficiency.

关键词： Measurement Multi-access edge computing Costs Numerical analysis Power control Information age Digital twins Synchronization Resource management Optimization

来源：评论

学校读者我要写书评

暂无评论

Integrating Existing Knowledge to Accelerate Buildings Renovation Rates in Europe 9th

Integrating Existing Knowledge to Accelerate Buildings Renov...

引用

9th International Conference on Decision Support System Technology, ICDSST 2023

作者： Karakosta, Charikleia Mylona, Zoi Papathanasiou, Jason Psarras, John Decision Support Systems Laboratory School of Electrical and Computer Engineering National Technical University of Athens 9 Iroon Polytechniou str. Zografou Athens15780 Greece HOLISTIC IKE 507 Mesogion Av.Ag. Paraskevi Athens15343 Greece Department of Business Administration University of Macedonia 156 Egnatia Street Thessaloniki54636 Greece

ISBN: (纸本)9783031325335

Nowadays, boosting the implementation of energy efficiency measures in buildings and subsequently, mainstreaming energy efficiency financing is of paramount importance for the European Union towards achieving its goal of carbon-neutrality by 2050. Unfortunately, statistics have shown that a lot of effort is needed to achieve the Europe’s targets, since energy efficiency is not yet considered as an attractive investment by the financial sector. The lack of expertise and knowledge, as well as the different perspective of project developers and financing institutions are some indicative challenges that have to be overcome. Specific to energy efficiency in buildings this is reflected by the current insufficient trends observed in the renovation rates of buildings, which reveal the urgent need for action since this is the largest consumer of energy in Europe. Furthermore, a combination of public and private funding through innovative financing instruments is required to overcome current barriers that prevent mobilization of necessary investments. The aim of this paper is to set up a role-based methodological approach for the deployment of an integrated matchmaking mechanism on an ICT platform to boost energy efficiency investments in an easy-access and trust-worthy way. The methodology envisages to follow a multidisciplinary perspective which takes into account the interactions between various key factors, such as stakeholders and barriers, so as to facilitate the complex set of decision-making actions for building renovation. The core of concept centres around the definition of the roles of the potential users of a big data for buildings platform, their interdependency and requirements with the ultimate purpose of accelerating renovation rates. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

关键词： Energy efficiency

来源：评论

学校读者我要写书评

暂无评论

Hybrid Multi-Agent Deep Reinforcement Learning for Active-IRS-Based Rate Maximization Over 6G UAV Mobile Wireless Networks

Hybrid Multi-Agent Deep Reinforcement Learning for Active-IR...

引用

MILCOM, Military Communications Conference

作者： Shuming Yi Fei Wang Xi Zhang College of Electronic and Information Engineering Southwest University Chongqing China Department of Electrical and Computer Engineering Networking and Information Systems Laboratory Texas A&M University College Station TX USA

Intelligent reflecting surface (IRS) can significantly improve unmanned aerial vehicle (UAV) communication qualities by reconfiguring or even optimizing wireless propagation environments. For active IRS (AIRS) enabled UAV-based mobile wireless networks, we aim to maximize the sum transmission rates of all IoT devices (IoTDs), by jointly optimizing the AIRSs’ phase shifts and amplification factors and UAV trajectory under the discrete IRS model. First, we formulate a sum transmission rate maximization problem for IoTDs, where all IoTDs can transmit data simultaneously by integrating the non-orthogonal multiple access (NOMA) and frequency division multiple access (FDMA) techniques. The formulated problem is non-convex because the coupled phase shifts and amplification factors are both discrete variables based on the discrete IRS model. Hence, we develop multi-agent deep reinforcement learning (MADRL) based schemes by integrating the deep deterministic policy gradient (DDPG) and Dueling Deep Q Network (Dueling DQN) techniques to derive the UAV trajectory and the amplification factors and phase shifts of AIRSs, respectively. Finally, we validate and evaluate the performances improvements of our developed MADRL-based schemes through extensive simulations.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Prediction Aided by Surrogate Training

arXiv

引用

arXiv 2024年

作者： Xia, Eric Wainwright, Martin J. Laboratory for Information and Decision Systems Statistics and Data Science Center Department of Electrical Engineering & Computer Sciences Massachusetts Institute of Technology CambridgeMA United States Laboratory for Information and Decision Systems Statistics and Data Science Center Department of Mathematics Massachusetts Institute of Technology CambridgeMA United States

We study a class of prediction problems in which relatively few observations have associated responses, but all observations include both standard covariates as well as additional "helper" covariates. While the end goal is to make high-quality predictions using only the standard covariates, helper covariates can be exploited during training to improve prediction. Helper covariates arise in many applications, including forecasting in time series;incorporation of biased or mis-calibrated predictions from foundation models;and sharing information in transfer learning. We propose "prediction aided by surrogate training" (PAST), a class of methods that exploit labeled data to construct a response estimator based on both the standard and helper covariates;and then use the full dataset with pseudo-responses to train a predictor based only on standard covariates. We establish guarantees on the prediction error of this procedure, with the response estimator allowed to be constructed in an arbitrary way, and the final predictor fit by empirical risk minimization over an arbitrary function class. These upper bounds involve the risk associated with the oracle data set (all responses available), plus an overhead that measures the accuracy of the pseudo-responses. This theory characterizes both regimes in which PAST accuracy is comparable to the oracle accuracy, as well as more challenging regimes where it behaves poorly. We demonstrate its empirical performance across a range of applications, including forecasting of societal ills over time with future covariates as helpers;prediction of cardiovascular risk after heart attacks with prescription data as helpers;and diagnosing pneumonia from chest X-rays using machine-generated predictions as helpers. © 2024, CC BY.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Diff9D: Diffusion-Based Domain-Generalized Category-Level 9-DoF Object Pose Estimation

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2025年第7期47卷 5520-5537页

作者： Jian Liu Wei Sun Hui Yang Pengchao Deng Chongpei Liu Nicu Sebe Hossein Rahmani Ajmal Mian National Engineering Research Center for Robot Visual Perception and Control Technology College of Electrical and Information Engineering School of Robotics and the State Key Laboratory of Advanced Design and Manufacturing for Vehicle Body Hunan University Changsha China Institute of Artificial Intelligence and Robotics Xi’an Jiaotong University Xi’an China Department of Information Engineering and Computer Science University of Trento Trento Italy School of Computing and Communications Lancaster University Lancaster U.K. Department of Computer Science The University of Western Australia Crawley WA Australia

Nine-degrees-of-freedom (9-DoF) object pose and size estimation is crucial for enabling augmented reality and robotic manipulation. Category-level methods have received extensive research attention due to their potential for generalization to intra-class unknown objects. However, these methods require manual collection and labeling of large-scale real-world training data. To address this problem, we introduce a diffusion-based paradigm for domain-generalized category-level 9-DoF object pose estimation. Our motivation is to leverage the latent generalization ability of the diffusion model to address the domain generalization challenge in object pose estimation. This entails training the model exclusively on rendered synthetic data to achieve generalization to real-world scenes. We propose an effective diffusion model to redefine 9-DoF object pose estimation from a generative perspective. Our model does not require any 3D shape priors during training or inference. By employing the Denoising Diffusion Implicit Model, we demonstrate that the reverse diffusion process can be executed in as few as 3 steps, achieving near real-time performance. Finally, we design a robotic grasping system comprising both hardware and software components. Through comprehensive experiments on two benchmark datasets and the real-world robotic system, we show that our method achieves state-of-the-art domain generalization performance.

关键词： Shape Solid modeling Three-dimensional displays Pose estimation Training Robots Grasping Diffusion models Real-time systems Point cloud compression

来源：评论

学校读者我要写书评

暂无评论

Computing Power in the Sky: Digital Twin-assisted Collaborative Computing with Multi-UAV Networks

引用

IEEE Transactions on Vehicular Technology 2025年

作者： Wang, Chao Han, Yu Zhang, Long Jia, Ziye Zhang, Hongliang Hong, Choong Seon Han, Zhu Hebei University of Engineering School of Information and Electrical Engineering Handan056038 China Qilu University of Technology Shandong Academy of Sciences Key Laboratory of Computing Power Network and Information Security Ministry of Education Jinan250014 China Nanjing University of Aeronautics and Astronautics College of Electronic and Information Engineering Nanjing211106 China Southeast University National Mobile Communications Research Laboratory Jiangsu Nanjing211111 China Peking University State Key Laboratory of Advanced Optical Communication Systems and Networks School of Electronics Beijing100871 China Kyung Hee University Department of Computer Science and Engineering Gyeonggi-do Yongin-si17104 Korea Republic of University of Houston Department of Electrical and Computer Engineering HoustonTX77004 United States Kyung Hee University Department of Computer Science and Engineering Seoul446-701 Korea Republic of

The collaboration of computing powers (CPs) among unmanned aerial vehicles (UAVs)-mounted edge servers is essential to handle data-intensive tasks of user equipments (UEs). This paper presents a multi-UAV computing power network (CPN) that orchestrates the CP resources of edge servers to cooperatively process tasks through collaborative computing. Due to the frequent interactions among edge servers and the increased complexities for large-scale task executions, we propose a digital twin (DT)-driven multi-UAV CPN framework, where the aerial DT network is created as a digital replica of the multi-UAV network to assist collaborative computing. A joint optimization problem of the service association, offloading power control, task partitions, CP allocation, and UAV trajectory design is formulated to minimize the long-term average delay of UEs. We use the Lyapunov technique to tackle the long-term energy constraint, forming a tractable optimization problem, which is modeled as a Markov decision process to balance the UAV's deficit queue stability and the UE's task completion delay. A hybrid soft actor-critic based deep reinforcement learning algorithm is developed to optimize the joint design by sampling the hybrid action from the Gumbel and Gaussian distributions. Simulation results verify that the multi-UAV CPN contributes to significant performance gains when supported by DTs. © 2025 IEEE.

关键词： Markov processes

来源：评论

学校读者我要写书评

暂无评论

Optimal Power Control Based-Metaheuristic Algorithm Based Variable Speed Wind Energy System of DFIG

Optimal Power Control Based-Metaheuristic Algorithm Based Va...

引用

International Smart Grid Workshop and Certificate Program (ISGWCP)

作者： Boussaid Ibrahim Harrouz Abdelkader Merahi Farid Ibrahim Yaichi Ilhami Colak Korhan Kayisli Virgil Dumbrava Electrical Engineering Department Laboratory of sustainable development and computer science (LDDI) University of Adrar Adrar Algeria Laboratory of sustainable development and computer science (LDDI) University of Adrar Adrar Algeria Electrical Engineering Department Automatic Laboratory of Setif University of Ferhat Abbas Setif1 Setif Algeria Department of Electrical Electronics Eng. Faculty of Engineering Architecture Nisantasi University Istanbul Turkey Department of Electric Power Systems Polytechnic University of Bucharest Bucharest Romania

This article presents a novel method for controlling a doubly fed induction generator (DFIG) wind energy system in the presence of variable wind speeds. In order to achieve decoupling control of the active and reactive power, the rotor-side converter (RSC) of the DFIG is controlled using field-oriented control (FOC). The system is very complicated and nonlinear, making it challenging to adjust the PI gains appropriately. The present research proposes the utilization of the Grey Wolf Optimization (GWO) algorithm as an approach to optimize the gains of a Proportional-Integral (PI) regulator. The simulation results demonstrate that the GWO-PI provides a lower criterion value and an improved response than the conventional PI.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Robust Adaptive Control Allocation Schemes for Overactuated Underwater Vehicles under Actuator Faults

引用

IFAC-PapersOnLine 2022年第6期55卷 67-72页

作者： Waseem Akram Alessandro Casavola Nikola Miškovic Department of Computer Engineering Modeling Electronics and Systems (DIMES) University of Calabria Rende Italy Faculty of Electrical Engineering and Computing LABUST - Laboratory for Underwater Systems and Technologies University of Zagreb Unska 3 Zagreb Croatia

In this paper, we investigate the fault-tolerant path following control problem for an overactuated underwater vehicle subject to input saturation and rate of change constraints, actuator faults and parametric (mass and inertia) uncertainty. A robust adaptive control allocation strategy is proposed capable to reconfigure the distribution of the control effort among the remaining healthy actuators in the case some of them failed. The online estimation of actuators' effectiveness and bias fault parameters is integrated with the control allocation and reconfiguration unit. The robustness and performance of the scheme are shown by evaluating the results of simulations and pool experiments of different path-following control problems using overactuated underwater vehicles.

关键词： Adaptive control Underwater vehicles Actuator faults trajectory tracking

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：