Large-Scale Kernel Methods for Independence Testing

Abstract

Representations of probability measures in reproducing kernel Hilbert spaces provide a flexible framework for fully nonparametric hypothesis tests of independence, which can capture any type of departure from independence, including nonlinear associations and multivariate interactions. However, these approaches come with an at least quadratic computational cost in the number of observations, which can be prohibitive in many applications. Arguably, it is exactly in such large-scale datasets that capturing any type of dependence is of interest, so striking a favourable tradeoff between computational efficiency and test performance for kernel independence tests would have a direct impact on their applicability in practice. In this contribution, we provide an extensive study of the use of large-scale kernel approximations in the context of independence testing, contrasting block-based, Nystrom and random Fourier feature approaches. Through a variety of synthetic data experiments, it is demonstrated that our novel large scale methods give comparable performance with existing methods whilst using significantly less computation time and memory.

29 pages, 6 figures

References in corpus (6)

• Measuring and testing dependence by correlation of distances
• Brownian distance covariance
• Kernel-based Conditional Independence Test and Application in Causal Discovery
• Fast Two-Sample Testing with Analytic Representations of Probability Measures
• FastMMD: Ensemble of Circular Discrepancy for Efficient Two-Sample Test
• A Kernel Test for Three-Variable Interactions with Random Processes

Cited by in corpus (26)

• Gaussian Processes and Kernel Methods: A Review on Connections and Equivalences
• The Chi-Square Test of Distance Correlation
• A fast algorithm for computing distance correlation
• Discovering and Deciphering Relationships Across Disparate Data Modalities
• The Exact Equivalence of Distance and Kernel Methods for Hypothesis Testing
• Measuring Association on Topological Spaces Using Kernels and Geometric Graphs
• Differentiable Causal Discovery from Interventional Data
• Knowledge-Based Regularization in Generative Modeling
• Multidecadal Cycles of the Climatic Index: Sunspots that Affect North and Northeast of Brazil
• Measuring Dependence with Matrix-based Entropy Functional
• On Kernel Derivative Approximation with Random Fourier Features
• Network Dependence Testing via Diffusion Maps and Distance-Based Correlations
• Using Kernel-Based Statistical Distance to Study the Dynamics of Charged Particle Beams in Particle-Based Simulation Codes
• Biased bootstrap sampling for efficient two-sample testing
• Causal Mosaic: Cause-Effect Inference via Nonlinear ICA and Ensemble Method
• BayesIMP: Uncertainty Quantification for Causal Data Fusion
• Nonparametric Causal Feature Selection for Spatiotemporal Risk Mapping of Malaria Incidence in Madagascar
• More Powerful Selective Kernel Tests for Feature Selection
• Fast Causal Discovery by Approximate Kernel-based Generalized Score Functions with Linear Computational Complexity
• Kernel Hypothesis Testing with Set-valued Data
• Improved Concentration Bounds for Gaussian Quadratic Forms
• Bayesian Kernelised Test of (In)dependence with Mixed-type Variables
• Neural Estimation of Statistical Divergences
• Testing semiparametric model-equivalence hypotheses based on the characteristic function
• A Kernel Test for Causal Association via Noise Contrastive Backdoor Adjustment
• A Note on Simulation-Based Inference by Matching Random Features