Synthetic Data and Generative AI

Chapter 16: Perturbed lattice point process: alternative to GMM

Inference, nearest neighbor graph

Abstract

This chapter covers additional topics most relevant to modern machine learning, from my book “Stochastic Processes and Simulations: A Machine Learning Perspective” [20]. The purpose is to introduce you to a new type of stochastic point processes [Wiki] with applications to sensor data, chemistry, physics (crystallography in particular) and cellular networks: for instance, to optimize the locations of cell towers or Internet-of-Things (IOT) devices.

The processes in question are known as perturbed lattices, and referred to here as Poisson–binomial processes for reasons that will soon become obvious. It is different both from Poisson and binomial ...