December 3, 2025Open Access

The Convergent Indian Buffet Process

Key Points

The proposed CIBP shows a bounded number of expected latent features as object count increases, unlike traditional methods.
Analysis reveals that the expected number of features under CIBP follows a Poisson distribution with a converging mean.
This framework is characterized by hierarchical distribution and random measure representations while assessing factor models.
Implications suggest CIBP may enhance feature selection in high-dimensional datasets with sparsity issues.

Abstract

We propose a new Bayesian nonparametric prior for latent feature models, called the Convergent Indian Buffet Process (CIBP). We show that under the CIBP, the number of latent features is distributed as a Poisson distribution, with the mean monotonically increasing but converging to a certain value as the number of objects goes to infinity. That is, the expected number of features is bounded above even when the number of objects goes to infinity, unlike the standard Indian Buffet Process, under which the expected number of features increases with the number of objects. We provide two alternative representations of the CIBP based on a hierarchical distribution and a completely random measure, which are of independent interest. The proposed CIBP is assessed on a high-dimensional sparse factor model.

Read Full Paperexternally

AIに質問

Bookmark

View Full Paper