A characterization of product-form exchangeable feature probability functions

DOI10.1214/17-AAP1333zbMATH Open1410.60100arXiv1607.02066OpenAlexW2964030503MaRDI QIDQ1661557FDOQ1661557

Authors: Marco Battiston, Stefano Favaro, Daniel M. Roy, Yee Whye Teh 

Publication date: 16 August 2018

Published in: The Annals of Applied Probability (Search for Journal in Brave)

Abstract: We characterize the class of exchangeable feature allocations assigning probability $V_{n,k}pro{d}_{l=1}^k{W}_{m_l}{U}_{n-m_l}$ to a feature allocation of $n$ individuals, displaying $k$ features with counts $(m_1,ldots,m_k)$ for these features. Each element of this class is parametrized by a countable matrix $V$ and two sequences $U$ and $W$ of non-negative weights. Moreover, a consistency condition is imposed to guarantee that the distribution for feature allocations of $n-1$ individuals is recovered from that of $n$ individuals, when the last individual is integrated out. In Theorem 1.1, we prove that the only members of this class satisfying the consistency condition are mixtures of the Indian Buffet Process over its mass parameter $gamma$ and mixtures of the Beta--Bernoulli model over its dimensionality parameter $N$. Hence, we provide a characterization of these two models as the only, up to randomization of the parameters, consistent exchangeable feature allocations having the required product form.

Full work available at URL: https://arxiv.org/abs/1607.02066

Recommendations

• Feature allocations, probability functions, and paintboxes
• Gibbs-type Indian buffet processes
• Exchangeable trait allocations
• The Indian buffet process: an introduction and review
• A characterisation of Pólya tree distributions

zbMATH Keywords

Indian buffet processexchangeable feature allocationsGibbs-type partitions

Mathematics Subject Classification ID

Special processes (60K99) Exchangeability for stochastic processes (60G09)

Cites Work

• The Indian buffet process: an introduction and review
• Title not available (Why is that?)
• Exchangeable and partially exchangeable random partitions
• Combinatorial stochastic processes. Ecole d'Eté de Probabilités de Saint-Flour XXXII -- 2002.
• Feature allocations, probability functions, and paintboxes
• Title not available (Why is that?)
• Sufficient statistics and extreme points
• Central limit theorems for an Indian buffet model with random weights

Cited In (6)

• Feature allocations, probability functions, and paintboxes
• Gibbs-type Indian buffet processes
• Exchangeable trait allocations
• A simple proof of Pitman-Yor's Chinese restaurant process from its stick-breaking representation
• Sufficientness postulates for Gibbs-type priors and hierarchical generalizations
• Approximating predictive probabilities of Gibbs-type priors

This page was built for publication: A characterization of product-form exchangeable feature probability functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1661557)