Cross-City Latent Space Alignment for Consistency Region Embedding

Learning urban region embeddings has substantially advanced urban analysis, but their typical focus on individual cities leads to disparate embedding spaces, hindering cross-city knowledge transfer and the reuse of downstream task predictors. To tackle this issue, we present Consistent Region Embedding (CoRE), a unified framework integrating region embedding learning with cross-city latent space alignment. CoRE first embeds regions from two cities into separate latent spaces, followed by the alignment of latent space manifolds and fine-grained individual regions from both cities. This ensures compatible and comparable embeddings within aligned latent spaces, enabling predictions of various socioeconomic indicators without ground truth labels by migrating knowledge from label-rich cities. Extensive experiments show CoRE outperforms competitive baselines, confirming its effectiveness for cross-city knowledge transfer via aligned latent spaces.

Cities are complex systems with distinct regions (neighborhoods), each serving different functions like residential, commercial, or industrial areas. Understanding these urban regions is useful for tasks like population estimation or infrastructure planning. However, most AI models today analyze only one city at a time, making it hard to apply insights from one city to another.We developed Consistency Region Embedding (CoRE), a method that helps AI models learn and compare neighborhood patterns across multiple cities. Instead of treating each city separately, CoRE maps different cities into their own "concept spaces" and then aligns these spaces so that similar regions—even from different cities—are grouped together. For example, a business district in New York and one in Tokyo would have comparable representations, even if their exact layouts differ.This alignment allows urban planners and researchers to reuse task predictors trained on one city for another. Our experiments show that CoRE works better than existing methods, making cross-city knowledge sharing more effective. This could help improve decision-making in urban development, transportation, and public policy.