Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Heterogeneous Graph Neural Networks with Post-hoc Explanations for Multi-modal and Explainable Land Use Inference

Xuehao Zhai, Junqi Jiang, Adam Dejl, Antonio Rago, Fangce Guo, Francesca Toni, Aruna Sivakumar

TL;DR

An explainable framework for inferring land use that synergises heterogeneous graph neural networks (HGNs) with Explainable AI techniques, enhancing both accuracy and explainability is introduced.

Abstract

Urban land use inference is a critically important task that aids in city planning and policy-making. Recently, the increased use of sensor and location technologies has facilitated the collection of multi-modal mobility data, offering valuable insights into daily activity patterns. Many studies have adopted advanced data-driven techniques to explore the potential of these multi-modal mobility data in land use inference. However, existing studies often process samples independently, ignoring the spatial correlations among neighbouring objects and heterogeneity among different services. Furthermore, the inherently low interpretability of complex deep learning methods poses a significant barrier in urban planning, where transparency and extrapolability are crucial for making long-term policy decisions. To overcome these challenges, we introduce an explainable framework for inferring land use that synergises heterogeneous graph neural networks (HGNs) with Explainable AI techniques, enhancing both accuracy and explainability. The empirical experiments demonstrate that the proposed HGNs significantly outperform baseline graph neural networks for all six land-use indicators, especially in terms of 'office' and 'sustenance'. As explanations, we consider feature attribution and counterfactual explanations. The analysis of feature attribution explanations shows that the symmetrical nature of the `residence' and 'work' categories predicted by the framework aligns well with the commuter's 'work' and 'recreation' activities in London. The analysis of the counterfactual explanations reveals that variations in node features and types are primarily responsible for the differences observed between the predicted land use distribution and the ideal mixed state. These analyses demonstrate that the proposed HGNs can suitably support urban stakeholders in their urban planning and policy-making.

Heterogeneous Graph Neural Networks with Post-hoc Explanations for Multi-modal and Explainable Land Use Inference

TL;DR

An explainable framework for inferring land use that synergises heterogeneous graph neural networks (HGNs) with Explainable AI techniques, enhancing both accuracy and explainability is introduced.

Abstract

Urban land use inference is a critically important task that aids in city planning and policy-making. Recently, the increased use of sensor and location technologies has facilitated the collection of multi-modal mobility data, offering valuable insights into daily activity patterns. Many studies have adopted advanced data-driven techniques to explore the potential of these multi-modal mobility data in land use inference. However, existing studies often process samples independently, ignoring the spatial correlations among neighbouring objects and heterogeneity among different services. Furthermore, the inherently low interpretability of complex deep learning methods poses a significant barrier in urban planning, where transparency and extrapolability are crucial for making long-term policy decisions. To overcome these challenges, we introduce an explainable framework for inferring land use that synergises heterogeneous graph neural networks (HGNs) with Explainable AI techniques, enhancing both accuracy and explainability. The empirical experiments demonstrate that the proposed HGNs significantly outperform baseline graph neural networks for all six land-use indicators, especially in terms of 'office' and 'sustenance'. As explanations, we consider feature attribution and counterfactual explanations. The analysis of feature attribution explanations shows that the symmetrical nature of the `residence' and 'work' categories predicted by the framework aligns well with the commuter's 'work' and 'recreation' activities in London. The analysis of the counterfactual explanations reveals that variations in node features and types are primarily responsible for the differences observed between the predicted land use distribution and the ideal mixed state. These analyses demonstrate that the proposed HGNs can suitably support urban stakeholders in their urban planning and policy-making.
Paper Structure (27 sections, 17 equations, 11 figures, 6 tables)

This paper contains 27 sections, 17 equations, 11 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Land use inference
  4. Multi-modal data fusion
  5. Explainability
  6. Preliminaries
  7. Methodology
  8. Heterogeneous Attention Embedding
  9. Explainable AI methods for post-hoc explanations
  10. Feature attribution methods
  11. Counterfactual Explanations
  12. Empirical Experiments
  13. Dataset description
  14. Baseline models for comparison
  15. Measurement metrics
  16. ...and 12 more sections

Figures (11)

  • Figure 1: This figure depicts a heterogeneous mobility network featuring meta-relations that highlight various connection types. The left panel displays a central London area with nodes coloured by different types of mobility services, representing tube stations (blue), bus stops (green), and bike stations (orange). The middle panel presents a legend for road routes (solid black lines), tube routes (dashed blue lines), and bus routes (dashed green lines), accompanied by daily ridership graphs for each mode of transport . The right panel outlines meta-relations with nodes and their connections, illustrating the complexity of information in a multi-modal mobility network.
  • Figure 2: Overview of our framework for land use inference based on multi-mobility systems.
  • Figure 3: The heterogeneous input is based on three data sources: a) locations of the tube, bus, and cycling stations, marked in blue, green, and orange respectively; b) connectivity among tube stations in the tube network; c) road network at different levels, which is used to build neighbourhood connections.
  • Figure 4: Spatial distribution of residual of residence: GCN (left) and HGT (right)
  • Figure 5: $R^2$ of six land use indicators in four scenarios: (a) bus, bike, and tube; (b) bus and bike; (c) bus and tube; (d) bus
  • ...and 6 more figures

Theorems & Definitions (9)

  • Definition 1
  • Definition 2
  • Definition 3
  • Definition 4
  • Definition 5
  • Definition 6
  • Definition 7
  • Definition 8
  • Definition 9