Predicting Spatiotemporal Mobile Sensing-Based PM2.5 Concentrations Using Low-Rank Adapted Spatially Attentive Graph Neural Network

SummaryAI generated

The researchers proposed a spatial attention graph neural network method based on LoRA for predicting PM2.5 concentration data collected by mobile sensors. By combining the spatial attention mechanism with low-rank adaptation techniques, this method aims to improve the prediction accuracy of particulate matter concentration in terms of space and time. The research was published on the arXiv cs.AI preprint platform on September 7, 2026.

Related eventsRELATED EVENTS

2026-09-09 05:29OpenAI claims huge maths breakthrough on a famed ‘Millennium Problem’
2026-09-09 06:11OpenAI claims to have solved the Navier-Stokes equation problem, but its originality is questioned by the academic community
2026-09-08 22:00Google DeepMind Maps 9 Billion Possible DNA Variants

Key entitiesKEY ENTITIES

GujaratIndiaSurat

Coverage · reports per dayLANGUAGE SPLIT

Entity relations

SignalsSIGNALS

Keyword heat

Surat1
Gujarat1
India1

All reports (1)SOURCES

↗

A arXiv cs.AI en 2026-09-07 12:00

研究人员提出一种基于低秩适配空间注意力图神经网络（SA-GNN）的模型，用于预测印度古吉拉特邦苏拉特市的 PM2.5 浓度。该研究构建了包含气象变量、土地利用特征及 PM2.5 浓度的移动传感数据集，并采用均匀分割与 DBSCAN 聚类两种策略定义图节点。SA-GNN 结合特定簇 GRU 捕捉局部时间依赖，利用图注意力网络学习空间异质性，在性能上优于 LSTM、RNN、GRU 和 ANN 等基线模型。实验结果显示，该模型在测试集上达到 R² = 0.95，RMSE = 6.8，MAE = 4.2 μg/m³。该方法有效捕捉了城市空气质量快速波动与复杂空间交互，支持实时精细监测、个性化暴露追踪及及时健康预警。