相关参考文献
注意：仅列出部分参考文献，下载原文获取全部文献信息。
Article
Computer Science, Information Systems
Haoyu Ma et al.
Summary: This article investigates the scheduling framework of the federated multitask learning problem with a hard-cooperation structure over wireless networks. A dynamic user and task scheduling scheme with a block-wise incremental gradient aggregation algorithm is proposed, and a Lyapunov-drift-based scheduling scheme is further proposed. Simulation results show that the proposed scheme significantly reduces the communication latency compared to the state-of-the-art baseline schemes.
IEEE INTERNET OF THINGS JOURNAL
(2023)
Article
Telecommunications
Claudio Battiloro et al.
Summary: This paper proposes a novel dynamic resource allocation strategy for energy-efficient adaptive federated learning at the wireless network edge, with latency and learning performance guarantees. By utilizing Lyapunov stochastic optimization tools, the radio parameters and computation resources are dynamically optimized to achieve the best trade-off between power, latency, and performance of the federated learning task. The method is customized for federated least mean squares (LMS) estimation and federated training of deep convolutional neural networks, and numerical results demonstrate the effectiveness of the strategy.
IEEE TRANSACTIONS ON GREEN COMMUNICATIONS AND NETWORKING
(2023)
Article
Engineering, Electrical & Electronic
Roula Nassif et al.
Summary: In this article, the authors investigate decentralized optimization problems with individual cost functions and subspace constraints on minimizers. They propose an adaptive decentralized strategy that uses differential randomized quantizers to compress estimates before communication. The analysis shows that the strategy is stable in terms of mean-square error and average bit rate under certain conditions, allowing for small estimation errors with a low bit rate. Simulations confirm the effectiveness of the proposed approach.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Wanyi Ning et al.
Summary: In this paper, we propose GRSGD, a global renovating stochastic gradient descent method, which utilizes the previous global gradient to estimate the current global gradient and updates local gradients to accelerate distributed edge learning. The method effectively reduces communication overhead and brings the convergence direction closer to global optimization.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Yuxuan Sun et al.
Summary: This study considers an over-the-air federated edge learning (FEEL) system with analog gradient aggregation and proposes an energy-aware dynamic device scheduling algorithm to optimize the training performance within device energy constraints.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Xiaowen Cao et al.
Summary: The paper investigates transmission power control to combat against aggregation errors in Air-FEEL. It proposes a new power control design aiming at maximizing convergence speed by analyzing the convergence behavior of Air-FEEL subject to aggregation errors. Optimized power control is achieved to minimize optimality gaps under unbiased aggregation constraints, leading to significantly faster convergence compared to benchmark policies.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Mohammad Moltafet et al.
Summary: This article presents a status update system where multiple sensors transmit information about random processes to a sink. By optimizing the sampling action, transmit power allocation, and sub-channel assignment, the trade-off between power consumption and information freshness is achieved. A dynamic control algorithm and a sub-optimal solution are proposed and their performance is evaluated through numerical results.
IEEE TRANSACTIONS ON COMMUNICATIONS
(2022)
Article
Engineering, Civil
B. Ravi Kiran et al.
Summary: This paper summarizes deep reinforcement learning algorithms, provides a taxonomy of automated driving tasks, discusses key computational challenges in real world deployment of autonomous driving agents, and explores adjacent domains as well as the role of simulators in training agents.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Yu Zhang et al.
Summary: This paper provides a survey of Multi-Task Learning (MTL) from the perspective of algorithmic modeling, applications, and theoretical analyses. It discusses different MTL algorithms and their characteristics, as well as the combination of MTL with other learning paradigms. The paper also reviews MTL models for large-scale tasks or high-dimensional data, as well as dimensionality reduction and feature hashing. Real-world applications of MTL are examined, and theoretical analyses and future directions are discussed.
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING
(2022)
Article
Engineering, Electrical & Electronic
Wei Guo et al.
Summary: This paper investigates the joint device selection and power control scheme for wireless federated learning, proposing an AirComp-based adaptive reweighing scheme for model aggregation. The study includes a convergence analysis, upper bound derivation, and numerical results demonstrating close-to-optimal performance of the proposed wireless FL algorithm.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Kaizhong Tin et al.
Summary: In many real-world applications, data are distributed across different regions and come from different distributions, resulting in multiple learning tasks. Distributed multi-task relationship learning (DMTRL) improves the performance of each task by learning task relationships from the data. However, privacy protection is a challenge, and we propose the DRUPE method to address this.
2022 31ST INTERNATIONAL CONFERENCE ON COMPUTER COMMUNICATIONS AND NETWORKS (ICCCN 2022)
(2022)
Article
Engineering, Electrical & Electronic
Dongzhu Liu et al.
Summary: This paper investigates privacy-preserving federated learning systems with uncoded transmission and explores adaptive power allocation strategies to minimize the learning optimality gap under privacy and power constraints. Simulation results demonstrate the importance of dynamic power allocation and the potential advantages of non-orthogonal multiple access.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Mingzhe Chen et al.
Summary: This article discusses the challenges of training federated learning algorithms over a realistic wireless network and proposes an optimization model to minimize the FL loss function, providing a method to improve identification accuracy.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Wenqi Shi et al.
Summary: In this paper, a joint device scheduling and resource allocation policy is proposed to maximize model accuracy within a given total training time budget for latency constrained wireless FL. The accuracy maximization problem is decomposed into two sub-problems and solved accordingly. Experimental results demonstrate the superiority of the proposed policy under various settings.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Shuo Wan et al.
Summary: Federated learning (FL) has emerged as an important and promising learning scheme in IoT, allowing devices to jointly learn a model without sharing raw data. The decentralized nature of FL requires frequent model exchange through wireless networks, with limited bandwidth and package loss impacting interactions in training. Network scheduling plays a significant role in determining the convergence of FL.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2021)
Article
Computer Science, Information Systems
Chengxi Li et al.
Summary: This article investigates FL in an IoT environment and proposes two new FL algorithms based on CS. Experiments show that these algorithms outperform baseline algorithms.
IEEE INTERNET OF THINGS JOURNAL
(2021)
Proceedings Paper
Computer Science, Information Systems
Hua Tang et al.
Summary: This paper introduces a novel coded distributed multi-task learning scheme, which reduces the communication load in both uplink and downlink by utilizing local information and specific repetitive placement and computation. The proposed approach successfully decreases the communication cost for distributed MTL framework.
2021 IEEE GLOBAL COMMUNICATIONS CONFERENCE (GLOBECOM)
(2021)
Article
Engineering, Electrical & Electronic
Jie Xu et al.
Summary: This study investigates the long-term perspective of resource allocation in wireless federated learning networks, showing the significant impact of temporal client selection patterns on learning performance. Data-driven experiments were designed to support the findings and a new algorithm was proposed to achieve long-term performance guarantee.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2021)
Article
Engineering, Electrical & Electronic
Sarit Khirirat et al.
Summary: The paper discusses the shift in optimization algorithm operating regime in the era of big data and introduces the improvement in solution accuracy through compression error compensation. Through experiments, it shows the advantages of Hessian-aided error compensation in quadratic problems and the strong convergence guarantees for stochastic gradient descent.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Computer Science, Hardware & Architecture
Canh T. Dinh et al.
Summary: The paper introduces a Federated Learning algorithm called FEDL, which can handle heterogeneous data from mobile user equipment and is applied as a resource allocation optimization problem in wireless networks. Experimental results demonstrate that in various settings, FEDL outperforms the original FedAvg algorithm in terms of convergence rate and test accuracy.
IEEE-ACM TRANSACTIONS ON NETWORKING
(2021)
Article
Engineering, Electrical & Electronic
Nir Shlezinger et al.
Summary: Traditional deep learning models are trained at a centralized server with data samples collected from users, but Federated Learning (FL) offers a new approach to training models without sharing user data. By using tools from quantization theory, this study addresses the challenge of repeated transmission of trained models by proposing a universal vector quantization scheme for FL (UVeQFed). The combination of universal vector quantization methods with FL results in a decentralized training system with minimal distortion and improved accuracy in the aggregated model.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Qiang Zhou et al.
Article
Engineering, Electrical & Electronic
Guangxu Zhu et al.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2020)
Article
Engineering, Electrical & Electronic
Roula Nassif et al.
IEEE SIGNAL PROCESSING MAGAZINE
(2020)
Article
Engineering, Electrical & Electronic
Kai Yang et al.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2020)
Article
Engineering, Electrical & Electronic
Mohammad Mohammadi Amiri et al.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2020)
Article
Engineering, Electrical & Electronic
Yeongjin Kim et al.
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
(2019)
Article
Computer Science, Information Systems
Shuang Zhao et al.
IEEE INTERNET OF THINGS JOURNAL
(2018)
Article
Engineering, Electrical & Electronic
Kalpana Naidu et al.
IEEE TRANSACTIONS ON COMMUNICATIONS
(2016)
Article
Computer Science, Artificial Intelligence
Yangyang Xu et al.
SIAM JOURNAL ON IMAGING SCIENCES
(2013)
Article
Mathematics, Applied
Meisam Razaviyayn et al.
SIAM JOURNAL ON OPTIMIZATION
(2013)
Article
Engineering, Electrical & Electronic
MG Rabbat et al.
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS
(2005)
Article
Computer Science, Theory & Methods
MF Duarte et al.
JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING
(2004)
