| 顾及地理-语义动态的城市热点预测框架 |
| |
| 引用本文： | 沙恒宇,金广垠,程光权,黄金才,吴克宇.顾及地理-语义动态的城市热点预测框架[J].地球信息科学,2022,24(1):25-37. |
| |
| 作者姓名： | 沙恒宇 金广垠 程光权 黄金才 吴克宇 |
| |
| 作者单位： | 国防科技大学系统工程学院,长沙 410000 |
| |
| 基金项目： | 国家自然科学基金项目(62073333);国家自然科学基金青年科学基金项目(62001495);湖南省自然科学基金青年科学基金项目(2020JJ5675)。 |
| |
| 摘 要： | 城市热点时空预测是城市管理和智慧城市建设的一项长期而富有挑战性的任务。准确地进行城市热点时空预测可以提高城市规划、调度和安全保障能力并降低资源消耗。现有的区域级深度时空预测方法主要利用基于地理网格的图像、给定的网络结构或额外的数据来获取时空动态。通过从原始数据中挖掘出潜在的自语义信息,并将其与基于地理空间的网格图像融合,也可以提高时空预测的性能,基于此,本文提出了一种新的深度学习方法地理语义集成神经网络(GSEN),将地理预测神经网络和语义预测神经网络相叠加。GSEN模型综合了预测递归神经网络（PredRNN）、图卷积预测递归神经网络（GC-PredRNN）和集成层的结构,从不同的角度捕捉时空动态。并且该模型还可以与现实世界中一些潜在的高层动态进行关联,而不需要任何额外的数据。最终在3个不同领域的实际数据集上对本文提出的模型进行了评估,均取得了很好的预测效果,实验结果表明GSEN模型在不同城市热点时空预测任务中的推广性和有效性,利用该模型可以更好地进行城市热点时空预测,解决一系列如犯罪、火灾、网约车预订等等现代城市发展中亟需解决的相关问题。
|
| 关 键 词： | 时空预测 智慧城市 城市热点 语义建模 预测递归神经网络 图卷积神经网络 |
| 收稿时间： | 2021-05-04 |
A Deep Urban Hotspots Prediction Framework with Modeling Geography-Semantic Dynamics |
| |
| SHA Hengyu,JIN Guangyin,CHENG Guangquan,HUANG Jincai,WU Keyu.A Deep Urban Hotspots Prediction Framework with Modeling Geography-Semantic Dynamics[J].Geo-information Science,2022,24(1):25-37. |
| |
| Authors: | SHA Hengyu JIN Guangyin CHENG Guangquan HUANG Jincai WU Keyu |
| |
| Institution: | College of Systems Engineering, National University of Defense Technology, Changsha 410000, China |
| |
| Abstract: | Urban hotspots prediction is a basic but significant task for future urban management. Accurate urban hotspots prediction can improve the efficiency of urban planning and security construction. Existing deep learning methods mainly adopt geographic grid maps, provided urban network, or external data to capture spatiotemporal dynamics. However, we observe that mining some latent self-semantics from raw data and fusing them with geospatial based grid images can also improve the performance of spatiotemporal predictions. In this paper, we propose Geographic-Semantic Ensemble Neural Network (GSEN), a novel deep learning approach, to stack geographical prediction neural network and semantical prediction neutral network. GSEN model integrates the structures of Predictive Recurrent Neural Network (PredRNN), Graph Convolutional Predictive Recurrent Neural Network (GC-PredRNN), and Ensemble Layer to capture spatiotemporal dynamics from different views. Furthermore, this model can also be correlated with some latent high-level dynamics in the real-world without any external data. We evaluate our proposed model on three different real-world datasets. The experimental results demonstrate the generalization and effectiveness of GSEN in different urban hotspots spatiotemporal prediction tasks. |
| |
| Keywords: | spatiotemporal prediction smart city city hotspot semantic modeling predictive recurrent neural network graph convolutional neural network |
| 本文献已被 维普 等数据库收录！ |
| 点击此处可从《地球信息科学》浏览原始摘要信息 |
| 点击此处可从《地球信息科学》下载免费的PDF全文 |
|
