Fine-Grained Urban Flow Inference with Multi-scale Representation Learning

TL;DR

UrbanMSR addresses the FUFI problem by learning dynamic multi-scale geographic representations at both neighborhood and city scales and fusing them through private and interactive decoders, guided by self-supervised contrastive pretraining. It enforces the structural constraint that fine-grained flows aggregate to their coarse counterparts via $M^{2}$ normalization and introduces a discriminative loss to reduce redundancy between scales. The approach shows consistent improvements over state-of-the-art FUFI methods across three real-world datasets, validating the effectiveness of dynamic cross-scale learning and fusion for fine-grained urban flow inference. The work advances practical FUFI by enabling accurate fine-grained predictions from coarse data with scalable multi-scale representations, benefiting traffic management and urban planning.

Abstract

Fine-grained urban flow inference (FUFI) is a crucial transportation service aimed at improving traffic efficiency and safety. FUFI can infer fine-grained urban traffic flows based solely on observed coarse-grained data. However, most of existing methods focus on the influence of single-scale static geographic information on FUFI, neglecting the interactions and dynamic information between different-scale regions within the city. Different-scale geographical features can capture redundant information from the same spatial areas. In order to effectively learn multi-scale information across time and space, we propose an effective fine-grained urban flow inference model called UrbanMSR, which uses self-supervised contrastive learning to obtain dynamic multi-scale representations of neighborhood-level and city-level geographic information, and fuses multi-scale representations to improve fine-grained accuracy. The fusion of multi-scale representations enhances fine-grained. We validate the performance through extensive experiments on three real-world datasets. The resutls compared with state-of-the-art methods demonstrate the superiority of the proposed model.

Figures (2)

• Figure 1: Fine-grained urban flow inference.
• Figure 2: The overall structure of UrbanMSR consists of three main steps: Stage I and II focus on pre-training the neighborhood-level encoder and city-level encoder. Then, the pre-trained encoders are transferred to the model for fine-tuning (Stage III).