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

Prediction of taxi service demand and supply is essential for improvingcustomer's experience and provider's profit. Recently, graph neural networks(GNNs) have been shown promising for this application. This approach modelscity regions as nodes in a transportation graph and their relations as edges.GNNs utilize local node features and the graph structure in the prediction.However, more efficient forecasting can still be achieved by following two mainroutes; enlarging the scale of the transportation graph, and simultaneouslyexploiting different types of nodes and edges in the graphs. However, bothapproaches are challenged by the scalability of GNNs. An immediate remedy tothe scalability challenge is to decentralize the GNN operation. However, thiscreates excessive node-to-node communication. In this paper, we firstcharacterize the excessive communication needs for the decentralized GNNapproach. Then, we propose a semi-decentralized approach utilizing multiplecloudlets, moderately sized storage and computation devices, that can beintegrated with the cellular base stations. This approach minimizesinter-cloudlet communication thereby alleviating the communication overhead ofthe decentralized approach while promoting scalability due to cloudlet-leveldecentralization. Also, we propose a heterogeneous GNN-LSTM algorithm forimproved taxi-level demand and supply forecasting for handling dynamic taxigraphs where nodes are taxis. Extensive experiments over real data show theadvantage of the semi-decentralized approach as tested over our heterogeneousGNN-LSTM algorithm. Also, the proposed semi-decentralized GNN approach is shownto reduce the overall inference time by about an order of magnitude compared tocentralized and decentralized inference schemes.

Semi-decentralized Inference in Heterogeneous Graph Neural Networks for Traffic Demand Forecasting: An Edge-Computing Approach