PreDisM: Pre-Disaster Modelling With CNN Ensembles for At-Risk Communities
This work addresses a critical gap in disaster mitigation for at-risk communities, enabling better resource allocation by state actors and NGOs, though it is incremental in applying existing methods to a new domain.
The paper tackles the problem of predicting building-level damages before natural disasters occur, introducing PreDisM, an ensemble model that combines ResNets, fully connected layers, and decision trees to estimate structural weaknesses, achieving good performance and responsiveness across disaster types.
The machine learning community has recently had increased interest in the climate and disaster damage domain due to a marked increased occurrences of natural hazards (e.g., hurricanes, forest fires, floods, earthquakes). However, not enough attention has been devoted to mitigating probable destruction from impending natural hazards. We explore this crucial space by predicting building-level damages on a before-the-fact basis that would allow state actors and non-governmental organizations to be best equipped with resource distribution to minimize or preempt losses. We introduce PreDisM that employs an ensemble of ResNets and fully connected layers over decision trees to capture image-level and meta-level information to accurately estimate weakness of man-made structures to disaster-occurrences. Our model performs well and is responsive to tuning across types of disasters and highlights the space of preemptive hazard damage modelling.