Reproduce “Exact, Fast and Expressive Poisson Point Processes via Squared Neural Families” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Russell Tsuchida, Cheng Soon Ong, Dino Sejdinovic
The authors introduce squared neural Poisson point processes (SNEPPPs), which model event intensity as the squared length of neural-network outputs. When pairwise feature products have closed-form integrals, the method computes integrated intensity exactly, avoiding numerical integration for this likelihood term. In wildfire scaling, SNEPPP fit in under 8 hours; the naive method ran out of memory beyond 5.5%. This compares computation, not predictive quality.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.