A multi-layer GCN on a proximity graph of Chicago crime grid cells achieves 78% accuracy in crime type classification and hotspot prediction, outperforming KDE and SVM baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Crime Hotspot Prediction Using Deep Graph Convolutional Networks
A multi-layer GCN on a proximity graph of Chicago crime grid cells achieves 78% accuracy in crime type classification and hotspot prediction, outperforming KDE and SVM baselines.