A Novel Multiscale Framework for Testing Independence: Efficient Detection of Explicit or Implicit Functional Relationships

TL;DR

This paper introduces a multiscale framework for independence testing that effectively detects explicit or implicit functional relationships, improving power and efficiency over existing methods.

Contribution

It presents a novel multiscale testing framework and a new test method that enhance detection power and computational efficiency for independence testing.

Findings

The proposed tests outperform existing methods in simulated datasets.

The framework effectively detects various types of dependence.

A visualization method aids in localizing dependence within data.

Abstract

In this article, we consider the problem of testing the independence between two random variables. Our primary objective is to develop tests that are highly effective at detecting associations arising from explicit or implicit functional relationship between two variables. We adopt a multiscale approach by analyzing neighborhoods of varying sizes within the dataset and aggregating the results. We introduce a general testing framework designed to enhance the power of existing independence tests to achieve our objective. Additionally, we propose a novel test method that is powerful as well as computationally efficient. The performance of these tests is compared with existing methods using various simulated datasets. Additionally, a visualization method has been proposed for exploring the localization of dependence within datasets.