报告人:苏理云(重庆理工大学)
时间:2023年12月13日 16:00-
地点:数统学院LD402
摘要:Due to the spatial-temporal complexity of the traffic flow data, to improve the performance of modeling, interpretability, and prediction accuracy of the prediction model, we propose a novel Generalized Spatial-Temporal Regression Graph Convolutional Transformer with an Auto-correlation mechanism (Auto-GSTRGCT) that integrates generalized spatial-temporal regression and auto-correlation mechanisms into graph convolutional networks. The model decomposes the original spatial-temporal data feature extraction into the spatial plane and the temporal plane. On the spatial plane, a spatial weight network is used to learn the semantic spatial weights in the real environment. On the temporal plane, temporal correlations are extracted using an auto-correlation mechanism, and spatial dependencies are learned dynamically at the same time. Experiments on two real traffic flow datasets show that our model framework is capable of advancing the state-of-the-art.
简介:苏理云,重庆理工大学理学院统计与数据科学系主任、教授,2007年获四川大学概率论与数理统计博士学位,统计学硕士生导师,应用统计专业硕士生导师,MBA硕士生导师,丹麦奥尔堡大学访问学者,重庆数学学会常务理事。已在《人口研究》、《数理统计与管理》、《物理学报》、《Applied Soft Computing》、《Digital Signal Processing》、《Mechanical Systems and Signal Processing》、《Computers & Mathematics with Applications》等国内外重要期刊发表论文50余篇,其中SCI收录30余篇。主持国家社科基金、教育部人文社科基金、重庆市自然科学基金等省部级及以上项目10余项。2022年获重庆市社会科学优秀成果二等奖,主持重庆市公司产品改革重大项目1项,重庆市应用统计研究生联合培养基地负责人,重庆市一流课程《概率论》负责人,校级教学名师。
邀请人:李曼曼
欢迎广大师生积极参与!