My research topic is ‘Interpretable Spatio-Temporal Crime Prediction Models Based on Machine Learning’. This study aims to develop interpretable spatio-temporal crime prediction models in London. Current crime prediction models struggle to balance predictive accuracy and interpretability. Although machine learning models excel in processing multimodal data, their decision-making processes often lack transparency, raising concerns about fairness. Current models with interpretability often overlook the spatio-temporal dependencies in crime data and fail to address the interpretability aspects of deep learning models. Therefore, this study proposes a new approach incorporating spatio-temporal dependencies into machine learning and deep learning crime prediction algorithms and enhancing interpretability with post hoc and attention modules. Moreover, this study will incorporate multimodal data, such as mobility, social media, and street-view images, to enhance predictive performance and provide more comprehensive interpretations. Finally, this research will develop patrol route planning algorithms based on predicted crime hotspots and develop simulation algorithms to model their application in real scenarios. The research is expected to offer a holistic framework for interpretable crime prediction models and provide more reliable decision support for relevant departments.
–
Undergraduate Qualification: BSc Network Engineering, Anhui University
Postgraduate Qualification: MSc Urban Informatics, King’s College London
