Spatiotemporal Prediction of Theft Risk with Deep Inception-Residual Networks

Xinyue Ye,Lian Duan,Qiong Peng

doi:10.3390/smartcities4010013

Xinyue Ye, Lian Duan + Show 1 more

Open Access

https://doi.org/10.3390/smartcities4010013

Copy DOI

Abstract

Spatiotemporal prediction of crime is crucial for public safety and smart cities operation. As crime incidents are distributed sparsely across space and time, existing deep-learning methods constrained by coarse spatial scale offer only limited values in prediction of crime density. This paper proposes the use of deep inception-residual networks (DIRNet) to conduct fine-grained, theft-related crime prediction based on non-emergency service request data (311 events). Specifically, it outlines the employment of inception units comprising asymmetrical convolution layers to draw low-level spatiotemporal dependencies hidden in crime events and complaint records in the 311 dataset. Afterward, this paper details how residual units can be applied to capture high-level spatiotemporal features from low-level spatiotemporal dependencies for the final prediction. The effectiveness of the proposed DIRNet is evaluated based on theft-related crime data and 311 data in New York City from 2010 to 2015. The results confirm that the DIRNet obtains an average F1 of 71%, which is better than other prediction models.

Highlights

Safety is an important goal that smart cities should pursue
Spatiotemporal prediction of crimes helps law enforcement agencies identify the patterns relating to the proliferation of crime [4] to efficiently deploy limited police resources [5,6]
This paper proposes a novel deep learning model—a deformable image registration deep learning network derived from deep inception-residual networks (DIRNet)—that can extract the information of spatial and temporal crime dependence before fusing multiple datasets for prediction

Summary

Introduction

Safety is an important goal that smart cities should pursue. As Neirotti et al stated, the initiatives of smart cities are characterized by modern technology and aiming at improving the lives of the urban residents in various domains, such as development, safety, energy, etc. [1]. As Neirotti et al stated, the initiatives of smart cities are characterized by modern technology and aiming at improving the lives of the urban residents in various domains, such as development, safety, energy, etc. Spatiotemporal prediction of crimes helps law enforcement agencies identify the patterns relating to the proliferation of crime [4] to efficiently deploy limited police resources [5,6]. A variety of engineering techniques have recently been developed to characterize crime-related features in order to enhance the prediction power using Foursquare data [14], Twitter data [15], 911 events [16], and taxi trajectories [17]. The challenge is that the aforementioned methods require either extensive coverage of the relevant indicators—which often suffers from insufficient data availability—or heavy reliance on feature-engineering processes that tend to be cumbersome

Objectives

Results

Conclusion