Asynchronous Decentralized Learning of a Neural Network

Asynchronous Decentralized Learning over Unreliable Wireless Networks

Decentralized learning enables edge users to collaboratively train models by exchanging information via device-to-device communication, yet prior works have been limited to wireless networks with fixed topologies and reliable workers. In this work, we propose an asynchronous decentralized stochastic gradient descent (DSGD) algorithm, which is robust to the inherent computation and communication failures occurring at the wireless network edge. We theoretically analyze its performance and establish a non-asymptotic convergence guarantee. Experimental results corroborate our analysis, demonstrati…

Accurate, Energy-Efficient, Decentralized, Single-Hop, Asynchronous Time Synchronization Protocols for Wireless Sensor Networks

This paper concerns with the synchronization of infrastructure impoverished sensor networks under harsh conditions. It suggests three novel asynchronous, decentralized, energyefficient time synchronization protocols. The protocols require only single hop, sparse communication with unlabeled neighboring nodes of the network to determine accurately the time of the gateway node. The time of a node is considered as a dynamical variable of a discrete system whose evolution is asynchronously activated/inhibited by another dynamical switching system. The protocols are termed: Timed Sequential Asynchr…

规模鲁棒的及时异步去中心化学习

Scale-Robust Timely Asynchronous Decentralized Learning

本文研究异步去中心化学习系统，提出以陈旧性准则保证用户模型收敛，并证明在用户闲聊容量按 Ω(log n) 扩展时可在有限时间内收敛，而分布式机会主义方案需 Ω(n) 扩展才能保证有界陈旧性。

Learning Decentralized Wireless Resource Allocations with Graph Neural Networks

We consider the broad class of decentralized optimal resource allocation problems in wireless networks, which can be formulated as a constrained statistical learning problems with a localized information structure. We develop the use of Aggregation Graph Neural Networks (Agg-GNNs), which process a sequence of delayed and potentially asynchronous graph aggregated state information obtained locally at each transmitter from multi-hop neighbors. We further utilize model-free primal-dual learning methods to optimize performance subject to constraints in the presence of delay and asynchrony inherent…

Neural Network Equalization for Asynchronous Multitrack Detection in TDMR

The advent of multiple readers in magnetic recording opens the possibility of replacing the current industry's single-track detection with the more promising multitrack detection architectures. We have proposed a first solution, a generalized partial-response maximum-likelihood (GPRML) architecture, that extends the conventional PRML paradigm to jointly detect multiple asynchronous tracks. In this paper, we propose to replace the conventional communication-theoretic multiple-input multiple-output equalizer in the GPRML architecture with a neural network equalizer for better adaption to the non…

Synthesizing Decentralized Controllers with Graph Neural Networks and Imitation Learning

Dynamical systems consisting of a set of autonomous agents face the challenge of having to accomplish a global task, relying only on local information. While centralized controllers are readily available, they face limitations in terms of scalability and implementation, as they do not respect the distributed information structure imposed by the network system of agents. Given the difficulties in finding optimal decentralized controllers, we propose a novel framework using graph neural networks (GNNs) to \emph{learn} these controllers. GNNs are well-suited for the task since they are naturally …