Recurrent Flow Networks: A Recurrent Latent Variable Model for Density Modelling of Urban Mobility

TL;DR

This paper introduces Recurrent Flow Networks, a novel model combining latent variables and normalizing flows to better capture the complex spatio-temporal dynamics of urban mobility demand for improved predictive accuracy.

Contribution

The paper proposes Recurrent Flow Networks that explicitly disentangle temporal and spatial variability in urban mobility demand modeling, enhancing predictive capabilities.

Findings

Model effectively captures complex urban mobility patterns.

Disentangling variability improves distribution matching.

Empirical results show better alignment with urban topologies.

Abstract

Mobility-on-demand (MoD) systems represent a rapidly developing mode of transportation wherein travel requests are dynamically handled by a coordinated fleet of vehicles. Crucially, the efficiency of an MoD system highly depends on how well supply and demand distributions are aligned in spatio-temporal space (i.e., to satisfy user demand, cars have to be available in the correct place and at the desired time). To do so, we argue that predictive models should aim to explicitly disentangle between temporal} and spatial variability in the evolution of urban mobility demand. However, current approaches typically ignore this distinction by either treating both sources of variability jointly, or completely ignoring their presence in the first place. In this paper, we propose recurrent flow networks (RFN), where we explore the inclusion of (i) latent random variables in the hidden state of recurrent neural networks to model temporal variability, and (ii) normalizing flows to model the spatial distribution of mobility demand. We demonstrate how predictive models explicitly disentangling between spatial and temporal variability exhibit several desirable properties, and empirically show how this enables the generation of distributions matching potentially complex urban topologies.