The emerging concern about data privacy and security has motivated the proposal of federated learning, which allows nodes to only synchronize locally-trained models instead their own original data. Conventional learning architecture, inherited from parameter server design, relies on highly centralized topologies assumption large nodes-to-server bandwidths. However, in real-world scenarios network capacities between are uniformly distributed smaller than that a datacenter. It is great...

Recent years have witnessed a rapid growth of deep-network based services and applications. A practical critical problem thus has emerged: how to effectively deploy the deep neural network models such that they can be executed efficiently. Conventional cloud-based approaches usually run in data center servers, causing large latency because significant amount transferred from edge center. In this paper, we propose JALAD, joint accuracy- latency-aware execution framework, which decouples so...

Recent years witnessed an increasing research attention in deploying deep learning models on edge devices for inference. Due to limited capabilities and power constraints, it may be necessary distribute the inference workload across multiple devices. Existing mechanisms divided model with assumption that are constructed a chain of layers. In reality, however, modern more complex, involving directed acyclic graph (DAG) rather than layers.In this paper, we present EdgeFlow, new distributed...

The federated learning paradigm protects private data from explicit leakage, yet exposing the model weights still raises serious privacy concerns with well-known attacks, such as membership inference attacks. It has been acknowledged that mechanisms homomorphic encryption and differential can be adopted to provide a higher level of protection. However, these may incur formidable amount overhead reductions in training performance, which make them unlikely employed real-world applications. In...

The emerging concern about data privacy and security has motivated the proposal of federated learning. Federated learning allows computing nodes to only synchronize locally- trained models instead their original in distributed training. Conventional architecture, inherited from parameter server design, relies on highly centralized typologies large nodes-to-server bandwidths. However, real-world scenarios, network capacities between are uniformly smaller than that centers. As a result, how...

This paper examines the effects of local corruption on total factor productivity (TFP) manufacturing firms in China. The empirical analysis is based a novel dataset we developed China at various disaggregation levels. results using fixed and instrumental variable estimation methods suggest that has an economically statistically significant negative effect firm productivity. estimated economic cost found to be high; one standard deviation increase reduces TFP by around 3.8%. We also find...

Abstract We explore the asymmetric effects of institutional differences on bilateral foreign direct investment ( FDI ) flows conditional countries' development levels, previous experiences investors and trade relations. The empirical results using data from 134 countries during 1990–2009 suggest that create entry barriers for only in North–South South–North directions, more so former. Furthermore, Southern appear to have a comparative advantage institutionally different developing countries....

This paper examines local corruption's direct and spillover effects on city-level foreign investment (FDI) flows in China. The empirical analysis is based a new dataset corruption at various disaggregation levels. Using convex combination structure spatial econometrics, we decompose the overall effect of into an indirect derive six novel results. First, has economically statistically significant negative home city FDI inflows (the effect). Second, positive to competing cities Third, 16 times...

The rapid growth of rich multimedia data in today's Internet, especially video traffic, has challenged the content delivery networks (CDNs). Caching serves as an important means to reduce user access latency so enable faster downloads. Motivated by dynamic nature real-world edge traces, this paper introduces a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">provably well</i> online caching policy environments where: 1) popularity is highly...

Real-time deep learning inference serving systems often require prohibitive resources and diverse user requirements. The existing design of mainly focusing on computation resource efficiency, largely ignoring the trade-off between bandwidth in need. Sub-optimal utilization usually leads to huge cost waste. In this paper, we tackle dual challenge computation-bandwidth cost-effectiveness by proposing A <sup xmlns:mml="http://www.w3.org/1998/Math/MathML"...

Geothermal resources are one of the most valuable renewable energy sources because their stability, reliability, cleanliness, safety and abundant reserves. Efficient economical remote sensing GIS (Geographic Information System) technology has high practical value in geothermal exploration. However, different study areas have formation mechanisms. In process establishing model, which factors used for modeling how to quantify reasonably still problems be analyzed studied. Taking Hangjiahu...

Abstract What explains changes in export sophistication across firms and destinations? This paper studies the effects of institutional similarity firm heterogeneity on using revealed product mix firm‐level data from China establishes eight stylised facts. First, more sophisticated products to destinations with similar institutions. Second, positive effect is weaker for higher productivity firms. Third, whilst exports private, foreign, joint‐venture are sophisticated, they less sensitive than...

With the rapid development of Industrial Internet, Network technology has been widely used in industrial field. In order to meet complex and diverse transmission requirements environment, field networks generally consist different network protocols. However, since each its own communication standards, it is difficult achieve collaborative management. Thus, how realize centralized management efficient intercommunication among Networks become a significant challenge. view this challenge, paper...

Download This Paper Open PDF in Browser Add to My Library Share: Permalink Using these links will ensure access this page indefinitely Copy URL DOI

Abstract Sound waves are significantly influenced by boundaries during their propagation in certain environmental conditions. The extent of this impact is related to the complexity boundary, such as case slopes and seamounts. In these areas, sound may deviate from original paths, resulting three-dimensional effects. Recent experiments simulations have demonstrated that effects occur when propagate over seamounts South China Sea, leading larger acoustic shadow regions. However, there limited...

In the process of numerical simulation, Finite Element Analysis (FEA) is commonly used for product design optimization. The model has to be modified according its associated optimization result obtained based on analysis, and mesh needs regenerated analysis verify modification. order reduce time regeneration, one effective way directly edit modification requirements. this paper, we present a novel editing method FE models. approach achieves complex accurate models through CAD operations...

First, from the point of view road marking, this article has analyzed visual navigation constraints which are dependent on geometry and response capability driver. The driver's psychology greatest impact induced perception in face rapid to system road-induced. Magnetic induction technology is cooperative vehicle infrastructure (CVI), used track analyze lane, keeping cornering control development process lane changing control. Markers it application United States, Japan, China other...

Highway agencies have gathered traffic data for a wide range of engineering and management purposes since the early days. Traffic is foundation decision-making related to planning, design, operation highway transportation system. The British Columbia Ministry Transportation Infrastructure (BCMOTI) in Canada has collected information various earliest days province. This article describes BCMOTI collection practice, which includes review needs, monitoring procedures methodologies,...

Does the degree of finance dependence, defined as firms' reliance on external for regular production activities, determine exporters' heterogeneous responses to real exchange rate shocks? This paper develops a stylized model illustrate role dependence in shaping pricing decisions when bilateral fluctuates. The features distribution costs, endogenous markup, and firm heterogeneity. In model, can impact export way isomorphic productivity but opposite direction: firms with high have demand...