Automated radiology report generation (RRG) can ease radiologist workload, yet most existing systems produce a report in a single forward pass, with no mechanism to check a claim against the image or revisit a finding once stated. We present CogRad, a cognitively inspired multi-a…
Spatio-temporal point-process models must often generalise across space when local event histories are sparse. We study whether exogenous spatial context can compensate in such regimes. Using a fixed log-Gaussian Cox process backbone, we compare an event-only model with the same…
Spatiotemporal point processes (STPPs) model event data in continuous time and space, with applications in mobility, epidemiology, and public safety. Recent neural STPPs span expressive intensity models, conditional density models, continuous-time latent dynamics, normalizing-flo…