References & Citations
Statistics > Machine Learning
Title: Poisson intensity estimation with reproducing kernels
(Submitted on 27 Oct 2016 (v1), last revised 26 Jun 2017 (this version, v3))
Abstract: Despite the fundamental nature of the inhomogeneous Poisson process in the theory and application of stochastic processes, and its attractive generalizations (e.g. Cox process), few tractable nonparametric modeling approaches of intensity functions exist, especially when observed points lie in a high-dimensional space. In this paper we develop a new, computationally tractable Reproducing Kernel Hilbert Space (RKHS) formulation for the inhomogeneous Poisson process. We model the square root of the intensity as an RKHS function. Whereas RKHS models used in supervised learning rely on the so-called representer theorem, the form of the inhomogeneous Poisson process likelihood means that the representer theorem does not apply. However, we prove that the representer theorem does hold in an appropriately transformed RKHS, guaranteeing that the optimization of the penalized likelihood can be cast as a tractable finite-dimensional problem. The resulting approach is simple to implement, and readily scales to high dimensions and large-scale datasets.
Submission history
From: Seth Flaxman [view email][v1] Thu, 27 Oct 2016 05:39:03 GMT (355kb,D)
[v2] Mon, 17 Apr 2017 10:56:41 GMT (325kb,D)
[v3] Mon, 26 Jun 2017 14:24:21 GMT (268kb,D)
Link back to: arXiv, form interface, contact.