HawkesNest: A Multi-Axis Synthetic Benchmark for Spatiotemporal Pattern Complexity
Researchers have introduced HawkesNest, a new synthetic benchmark designed to evaluate spatiotemporal point process (STPP) models. Unlike real-world datasets, HawkesNest offers controlled complexity along four axes: space-time entanglement, background heterogeneity, cross-type interaction, and domain topology. This allows for diagnostic stress tests of STPP models by isolating specific structural difficulties. Initial tests show that existing Hawkes-family baselines and neural models like AutoSTPP degrade under certain complexity increases, highlighting their sensitivities. AI
IMPACT Provides a new diagnostic tool for evaluating the robustness of spatiotemporal AI models.