New multi-modal graph network improves urban spatiotemporal forecasting

This paper introduces a novel approach for urban spatiotemporal forecasting by treating region-wise relationships as a multi-modal machine learning problem. The proposed method utilizes multi-graph convolution networks and incorporates two interaction techniques to enhance feature extraction and reduce generalization errors. Specifically, it employs grouped GCN for lower layers to integrate connectivity from different modalities and multi-linear relationship GCN for higher layers to learn more generalized features, outperforming existing baselines in prediction accuracy and efficiency on ridehailing demand forecasting tasks. AI

IMPACT Introduces novel multi-modal graph interaction techniques for improved accuracy and efficiency in urban spatiotemporal forecasting tasks.

RANK_REASON The item is an academic paper detailing a new method for spatiotemporal forecasting. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• Graph convolution network with similarity awareness and adaptive calibration for disease-induced deterioration prediction
• grouped GCN
• multi-graph convolution networks
• multi-linear relationship GCN
• multi-linear relationship networks
• Multi-Modal Graph Interaction for Multi-Graph Convolution Network in Urban Spatiotemporal Forecasting
• multi-modal machine learning
• ridehailing demand forecasting
• Urban computing
• Xu Geng

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →