DIET: Conditional independence testing with marginal dependence measures of residual information
This addresses a computational bottleneck for researchers and practitioners in statistics and machine learning performing conditional independence testing, offering a more efficient method without sacrificing power, though it appears incremental as it builds on existing CRT frameworks.
The paper tackles the computational intractability of conditional randomization tests (CRTs) for assessing variable predictiveness, proposing the decoupled independence test (DIET) which uses marginal independence statistics on information residuals to test conditional independence, and shows it achieves higher power than other tractable CRTs on synthetic and real benchmarks.
Conditional randomization tests (CRTs) assess whether a variable $x$ is predictive of another variable $y$, having observed covariates $z$. CRTs require fitting a large number of predictive models, which is often computationally intractable. Existing solutions to reduce the cost of CRTs typically split the dataset into a train and test portion, or rely on heuristics for interactions, both of which lead to a loss in power. We propose the decoupled independence test (DIET), an algorithm that avoids both of these issues by leveraging marginal independence statistics to test conditional independence relationships. DIET tests the marginal independence of two random variables: $F(x \mid z)$ and $F(y \mid z)$ where $F(\cdot \mid z)$ is a conditional cumulative distribution function (CDF). These variables are termed "information residuals." We give sufficient conditions for DIET to achieve finite sample type-1 error control and power greater than the type-1 error rate. We then prove that when using the mutual information between the information residuals as a test statistic, DIET yields the most powerful conditionally valid test. Finally, we show DIET achieves higher power than other tractable CRTs on several synthetic and real benchmarks.