# Importance conditional sampling for Pitman-Yor mixtures

## TL;DR

This paper introduces importance conditional sampling (ICS), a new efficient and stable method for Pitman-Yor mixture models, enhancing density estimation and clustering with better parameter robustness and extendability.

## Contribution

The paper proposes ICS, a novel sampling strategy for Pitman-Yor mixtures that improves efficiency, interpretability, and stability over existing methods, and extends to other complex models.

## Key findings

- ICS shows stable performance across various parameter settings.
- The method is computationally efficient and parallelizable.
- ICS can be extended to other nonparametric mixture models.

## Abstract

Nonparametric mixture models based on the Pitman-Yor process represent a flexible tool for density estimation and clustering. Natural generalization of the popular class of Dirichlet process mixture models, they allow for more robust inference on the number of components characterizing the distribution of the data. We propose a new sampling strategy for such models, named importance conditional sampling (ICS), which combines appealing properties of existing methods, including easy interpretability and a within-iteration parallelizable structure. An extensive simulation study highlights the efficiency of the proposed method which, unlike other conditional samplers, shows stable performances for different specifications of the parameters characterizing the Pitman-Yor process. We further show that the ICS approach can be naturally extended to other classes of computationally demanding models, such as nonparametric mixture models for partially exchangeable data.