Neural network speeds up estimating hidden event rates from repeated data
Amortized Variational Inference of Latent Intensities from Replicated Point Patterns
Machine Learning
Summary
Figuring out the hidden patterns behind events happening over time can be slow when done separately each time. The authors designed a neural network that learns from many examples to quickly estimate these hidden patterns all at once, even when the number of examples changes. This method doesn’t need to know the exact patterns beforehand and works well on both made-up data and real bike-sharing usage records. It speeds up the process compared to older methods while keeping accuracy high.
What this means in practice
- •For urban mobility planners: Estimate underlying bike share demand patterns faster from repeated trip data to improve resource allocation.
- •For online event platform operators: Infer user activity intensity quickly from multiple event logs to optimize system performance.
Authors
Wenqing Liu, Xenia Miscouridou, Déborah Sulem
Abstract
Bayesian inference for latent point process intensities often relies on per-dataset sampling or optimization, making repeated inference computationally expensive. We propose a neural-network-based amortized variational framework that learns a reusable inference rule mapping a variable number of replicated point patterns to a distribution over their shared latent intensity. Trained from event times using the point process likelihood, the model learns an amortized inference rule without ground-truth intensity supervision. At test time, it performs inference of the latent distribution in a forward pass. On synthetic data, additional realizations improve intensity recovery and yield narrower credible intervals with broadly stable coverage. Our method is tested on bicycle-sharing data and achieves the highest mean held-out log likelihood among the evaluated methods while requiring substantially less inference time than the Log-Gaussian Cox Process baseline.