Federated learning enables resource-constrained edge compute devices, such as mobile phones and IoT devices, to learn a shared model for prediction, while keeping the training data local. This decentralized approach to train models provides privacy, security, regulatory and economic benefits. In this work, we focus on the statistical challenge of federated learning when local data is non-IID. We first show that the accuracy of federated learning reduces significantly, by up to ~55% for neural networks trained for highly skewed non-IID data, where each client device trains only on a single class of data. We further show that this accuracy reduction can be explained by the weight divergence, which can be quantified by the earth mover's distance (EMD) between the distribution over classes on each device and the population distribution. As a solution, we propose a strategy to improve training on non-IID data by creating a small subset of data which is globally shared between all the edge devices. Experiments show that accuracy can be increased by ~30% for the CIFAR-10 dataset with only 5% globally shared data.
Blockchain technology can solve the problem of trust in the open network in a decentralized way. It has broad application prospects and has attracted extensive attention from academia and industry. The blockchain consensus algorithm ensures that the nodes in the chain reach consensus in the complex network environment, and the node status ultimately remains the same. The consensus algorithm is one of the core technologies of blockchain and plays a pivotal role in the research of blockchain technology. This article gives the basic concepts of the blockchain, summarizes the key technologies of the blockchain, especially focuses on the research of the blockchain consensus algorithm, expounds the general principles of the consensus process, and classifies the mainstream consensus algorithms. Then, focusing on the improvement of consensus algorithm performance, it reviews the research progress of consensus algorithms in detail, analyzes and compares the characteristics, suitable scenarios, and possible shortcomings of different consensus algorithms, and based on this, studies the future development trend of consensus algorithms for reference.
This paper studies the modelling and stabilization problem for a class of networked control systems (NCSs) with random time delays and packed dropouts based on the average dwell time (ADT) switching approach. A new effective sampling method is presented to deal with time-varying delays, which finally reduces the high dimension and complexity of the NCS model. A class of subsystems is developed for NCS based on a set of switching signals triggered by discretized time delays and packet dropouts. By constructing multiple quadratic Lyapunov-like functions that allowed both decreasing and increasing, sufficient switched stability conditions are derived and a set of mode-dependent feedback controllers are designed for each active mode based on the ADT switching control method to guarantee the stability of the proposed switched NCS with admissible ADT. A do-loop linear matrix inequalities (LMI) optimization problem is formulated to find the control law without predefining the switching sequence and the times when network-induced factors occur. A numerical example is given to illustrate the effectiveness of the proposed method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.