ÂÌñÉç

Join Zoom Meeting
(ID: 96298588872, passcode: 791527)

Join by phone
(US) +1 301-715-8592 (passcode: 791527)

Joining instructions:

Meeting host: uergash1@binghamton.edu

Join Zoom Meeting:

Presenter: Yuhang Liu
Advisor: Yingxue Zhang
When: Monday, February 24, 1pm

Where: EB T01 

Title:  Spatial temporal urban dynamics prediction: challenges, methods, and future directions.

Abstract: Urban dynamics prediction in traffic systems involves forecasting human mobility patterns, such as traffic speed and travel demand, using historical spatial temporal data collected from various IoT devices mounted on vehicles and public transport systems. These dynamics encompass a range of interconnected processes, including traffic flow, public transit utilization, and energy consumption, all of which evolve over time and space. Accurate prediction is essential for optimizing city operations, reducing congestion, and improving sustainability. However, the intricate spatial temporal dependencies and strong interconnections among different urban factors present significant modeling challenges. In this talk, I will begin with an overview of urban dynamics prediction, followed by a discussion of key challenges, such as capturing complex spatial temporal relationships and modeling the interdependence of various dynamic processes. Following this, I will review recent advancements in three major research directions: graph-based methods, generative models, and pre-trained models. Finally, I will discuss my recent research in this field and explore potential future directions.