English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Edge intelligence has arisen as a promising computing paradigm for supportingmiscellaneous smart applications that rely on machine learning techniques.While the community has extensively investigated multi-tier edge deployment fortraditional deep learning models (e.g. CNNs, RNNs), the emerging Graph NeuralNetworks (GNNs) are still under exploration, presenting a stark disparity toits broad edge adoptions such as traffic flow forecasting and location-basedsocial recommendation. To bridge this gap, this paper formally studies the costoptimization for distributed GNN processing over a multi-tier heterogeneousedge network. We build a comprehensive modeling framework that can capture avariety of different cost factors, based on which we formulate a cost-efficientgraph layout optimization problem that is proved to be NP-hard. Instead oftrivially applying traditional data placement wisdom, we theoretically revealthe structural property of quadratic submodularity implicated in GNN's uniquecomputing pattern, which motivates our design of an efficient iterativesolution exploiting graph cuts. Rigorous analysis shows that it providesparameterized constant approximation ratio, guaranteed convergence, and exactfeasibility. To tackle potential graph topological evolution in GNN processing,we further devise an incremental update strategy and an adaptive schedulingalgorithm for lightweight dynamic layout optimization. Evaluations withreal-world datasets and various GNN benchmarks demonstrate that our approachachieves superior performance over de facto baselines with more than 95.8% costeduction in a fast convergence speed.

GNN at the Edge: Cost-Efficient Graph Neural Network Processing over Distributed Edge Servers