English

Inferring the finest pattern of mutual independence from data

Machine Learning 2024-09-10 v1 Machine Learning Statistics Theory Methodology Statistics Theory

Abstract

For a random variable XX, we are interested in the blind extraction of its finest mutual independence pattern μ(X)\mu ( X ). We introduce a specific kind of independence that we call dichotomic. If Δ(X)\Delta ( X ) stands for the set of all patterns of dichotomic independence that hold for XX, we show that μ(X)\mu ( X ) can be obtained as the intersection of all elements of Δ(X)\Delta ( X ). We then propose a method to estimate Δ(X)\Delta ( X ) when the data are independent and identically (i.i.d.) realizations of a multivariate normal distribution. If Δ^(X)\hat{\Delta} ( X ) is the estimated set of valid patterns of dichotomic independence, we estimate μ(X)\mu ( X ) as the intersection of all patterns of Δ^(X)\hat{\Delta} ( X ). The method is tested on simulated data, showing its advantages and limits. We also consider an application to a toy example as well as to experimental data.

Keywords

Cite

@article{arxiv.2306.12984,
  title  = {Inferring the finest pattern of mutual independence from data},
  author = {G. Marrelec and A. Giron},
  journal= {arXiv preprint arXiv:2306.12984},
  year   = {2024}
}
R2 v1 2026-06-28T11:12:04.547Z