Learning A Universal Crime Predictor with Knowledge-guided Hypernetworks

TL;DR

HYSTL addresses cross-city crime prediction under semantic misalignment and heterogeneous crime dynamics by coupling a CrimeKG with an adaptive hypernetwork that generates crime-type-specific parameters for a shared spatio-temporal predictor. The framework builds a CrimeKG from Wikipedia metadata, uses metapath2vec++ embeddings as inputs to a hypernetwork, and plug-ins these embeddings into an A3TGCN-based backbone with attention-based temporal aggregation. Empirical results on NYC and Chicago demonstrate state-of-the-art performance, with substantial MAE/MAPE improvements and strong transferability across backbone models. The approach improves robustness to data sparsity and non-overlapping crime types, and points to future extensions in zero-shot/few-shot settings and interpretability for decision-makers. The combination of semantically informed embeddings and dynamic parameter generation offers a scalable path to unified, cross-city crime forecasting with practical public safety impact.

Abstract

Predicting crimes in urban environments is crucial for public safety, yet existing prediction methods often struggle to align the knowledge across diverse cities that vary dramatically in data availability of specific crime types. We propose HYpernetwork-enhanced Spatial Temporal Learning (HYSTL), a framework that can effectively train a unified, stronger crime predictor without assuming identical crime types in different cities' records. In HYSTL, instead of parameterising a dedicated predictor per crime type, a hypernetwork is designed to dynamically generate parameters for the prediction function conditioned on the crime type of interest. To bridge the semantic gap between different crime types, a structured crime knowledge graph is built, where the learned representations of crimes are used as the input to the hypernetwork to facilitate parameter generation. As such, when making predictions for each crime type, the predictor is additionally guided by its intricate association with other relevant crime types. Extensive experiments are performed on two cities with non-overlapping crime types, and the results demonstrate HYSTL outperforms state-of-the-art baselines.

Figures (3)

• Figure 1: The framework combines static temporal graphs representing crime data with embeddings from CrimeKG to enhance prediction. Each crime type generates specific parameters using a hypernetwork. ST-GNN Model extracts the temporal-spatial features, which are aggregated into context vectors for crime type-specific predictions.
• Figure 2: Ablation study of HYSTL framework, in terms of MAE and MAPE.
• Figure 3: Impact study for various hyperparameters in HYSTL’s performance on Chicago and New York crime data, in terms of MAE and MAPE.