Heterogeneity across clients in federated learning (FL) usually hinders the optimization convergence and generalization performance when aggregation of clients' knowledge occurs gradient space. For example, may differ terms data distribution, network latency, input/output space, and/or model architecture, which can easily lead to misalignment their local gradients. To improve tolerance heterogeneity, we propose a novel prototype (FedProto) framework server communicate abstract class...

The Internet of Things (IoT) extends the connectivity into billions IoT devices around world, where collect and share information to reflect status physical world. Autonomous Control System (ACS), on other hand, performs control functions systems without external intervention over an extended period time. integration ACS results in a new concept - autonomous (AIoT). sensors system status, based which intelligent agents as well Edge/Fog/Cloud servers make decisions for actuators react. In...

Graph neural networks (GNNs) have shown their superiority in modeling graph data. Owing to the advantages of federated learning, learning (FGL) enables clients train strong GNN models a distributed manner without sharing private A core challenge systems is non-IID problem, which also widely exists real-world For example, local data may come from diverse datasets or even domains, e.g., social and molecules, increasing difficulty for FGL methods capture commonly shared knowledge learn...

Microgrids (MGs) are small, local power grids that can operate independently from the larger utility grid. Combined with Internet of Things (IoT), a smart MG leverage sensory data and machine learning techniques for intelligent energy management. This article focuses on deep reinforcement (DRL)-based dispatch IoT-driven isolated MGs diesel generators (DGs), photovoltaic (PV) panels, battery. A finite-horizon partial observable Markov decision process (POMDP) model is formulated solved by...

Abstract Federated learning is a new paradigm that decouples data collection and model training via multi-party computation aggregation.As flexible setting, federated has the potential to integrate with other frameworks.We conduct focused survey of in conjunction algorithms. Specifically, we explore various algorithms improve vanilla averaging algorithm review fusion methods such as adaptive aggregation, regularization, clustered methods, Bayesian methods. Following emerging trends, also...

Anomaly detection for non-linear dynamical system plays an important role in ensuring the stability. However, it is usually complex and has to be solved by large-scale simulation which requires extensive computing resources. In this paper, we propose a novel anomaly scheme based on Long Short-Term Memory (LSTM) capture temporal changes of time sequence make multi-step predictions. Specifically, first present framework LSTM-based system, including data preprocessing, prediction detection....

Federated learning is a new paradigm that decouples data collection and model training via multi-party computation aggregation. As flexible setting, federated has the potential to integrate with other frameworks. We conduct focused survey of in conjunction algorithms. Specifically, we explore various algorithms improve vanilla averaging algorithm review fusion methods such as adaptive aggregation, regularization, clustered methods, Bayesian methods. Following emerging trends, also discuss...

The emerging cloud computing service market aims at delivering resources as a utility over the Internet with high quality. It has evolving unknown demand that is typically highly uncertain. Traditional provisioning methods either make idealized assumption of distribution or rely on extensive offline statistical analysis historical data. In this paper, we present an online adaptive learning approach to address optimal resource problem. Based stochastic loss model services, formulate problem...

column Share on Provisioning for large scale loss network systems with applications in cloud computing Authors: Yue Tan The Ohio State University UniversityView Profile , Yingdong Lu IBM Thomas J. Watson Research Center CenterView Cathy H. Xia Authors Info & Claims ACM SIGMETRICS Performance Evaluation ReviewVolume 40Issue 3December 2012 pp 83–85https://doi.org/10.1145/2425248.2425270Online:04 January 2012Publication History 8citation12DownloadsMetricsTotal Citations8Total Downloads12Last 12...

Resource provisioning, the task of planning sufficient amounts resources to meet service level agreements, has become an important management in emerging cloud computing services. In this paper, we present a stochastic modeling approach guide resource provisioning for future clouds as demand grows large. We focus on on-demand services and consider availability key quality constraint. A specific scenario under consideration is when can be measured base instances. develop asymptotic...

We study the accuracy of a scaled Poisson approximation to weighted sum independent random variables, focusing on in particular relative error tail distribution. establish moderate deviation bound using modified Stein-Chen method. Numerical experiments are also presented demonstrate quality approximation.

Microgrids (MGs) are small, local power grids that can operate independently from the larger utility grid. Combined with Internet of Things (IoT), a smart MG leverage sensory data and machine learning techniques for intelligent energy management. This paper focuses on deep reinforcement (DRL)-based dispatch IoT-driven isolated MGs diesel generators (DGs), photovoltaic (PV) panels, battery. A finite-horizon Partial Observable Markov Decision Process (POMDP) model is formulated solved by...

The Internet of Things (IoT) extends the connectivity into billions IoT devices around world, where collect and share information to reflect status physical world. Autonomous Control System (ACS), on other hand, performs control functions systems without external intervention over an extended period time. integration ACS results in a new concept - autonomous (AIoT). sensors system status, based which intelligent agents as well Edge/Fog/Cloud servers make decisions for actuators react. In...

We develop a data-driven adaptive control framework to password management in cyber security systems. A policy is the frontline of protection against attacks, which contains set rules on length, duration, etc. assume has censored lifetime, and maker determines duration without complete knowledge its true lifetime distribution. gradient based algorithm integrated with Bayesian learning framework. show that our converges optimal solution adapts non-stationary data.

Traditional federated learning (FL) methods often rely on fixed weighting for parameter aggregation, neglecting the mutual influence by others. Hence, their effectiveness in heterogeneous data contexts is limited. To address this problem, we propose an influence-oriented framework, namely FedC^2I, which quantitatively measures Client-level and Class-level Influence to realize adaptive aggregation each client. Our core idea explicitly model inter-client within FL system via well-crafted...

This paper addresses operation strategies of the physical showroom under e-commerce and studies joint decisions service pricing based on product quality. First, we propose an analytical model to capture feature consumer purchase behaviors examine three strategies: cooperation (CO), manufacturer (MO), e-retailer (RO). Then, equilibriums in above are obtained analyzed Besides, extend scenario endogenous The results as follows. contrast conventional wisdom, find that optimal may decrease with...

Event-centered information integration is regarded as one of the most pressing issues in improving disaster emergency management. Ontology plays an increasingly important role integration, and provides possibility for reasoning. However, development event ontology a laborious difficult task due to scale complexity emergencies. pattern modeling solution solve recurrent design problem, which can improve efficiency by reusing patterns. By study on characteristics numerous emergencies, this...