A Kernel Test of Goodness of Fit

Abstract

We propose a nonparametric statistical test for goodness-of-fit: given a set of samples, the test determines how likely it is that these were generated from a target density function. The measure of goodness-of-fit is a divergence constructed via Stein's method using functions from a Reproducing Kernel Hilbert Space. Our test statistic is based on an empirical estimate of this divergence, taking the form of a V-statistic in terms of the log gradients of the target density and the kernel. We derive a statistical test, both for i.i.d. and non-i.i.d. samples, where we estimate the null distribution quantiles using a wild bootstrap procedure. We apply our test to quantifying convergence of approximate Markov Chain Monte Carlo methods, statistical model criticism, and evaluating quality of fit vs model complexity in nonparametric density estimation.

14 pages, 9 figures

References in corpus (9)

• Equivalence of distance-based and RKHS-based statistics in hypothesis testing
• Austerity in MCMC Land: Cutting the Metropolis-Hastings Budget
• Measuring Sample Quality with Stein's Method
• Density Estimation in Infinite Dimensional Exponential Families
• Degenerate U- and V-statistics under weak dependence: Asymptotic theory and bootstrap consistency
• Gradient-free Hamiltonian Monte Carlo with Efficient Kernel Exponential Families
• A Wild Bootstrap for Degenerate Kernel Tests
• A Kernelized Stein Discrepancy for Goodness-of-fit Tests and Model Evaluation
• Control functionals for Monte Carlo integration

Cited by in corpus (64)

• Kernel Mean Embedding of Distributions: A Review and Beyond
• Learning Decentralized Controllers for Robot Swarms with Graph Neural Networks
• Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm
• Measuring Sample Quality with Kernels
• Detecting Out-of-Distribution Inputs to Deep Generative Models Using Typicality
• Gradient Estimators for Implicit Models
• A Non-Asymptotic Analysis for Stein Variational Gradient Descent
• Fixed point characterizations of continuous univariate probability distributions and their applications
• A Measure-Theoretic Approach to Kernel Conditional Mean Embeddings
• Distributionally Robust Bayesian Optimization
• Stochastic Runge-Kutta Accelerates Langevin Monte Carlo and Beyond
• Benchmarking Simulation-Based Inference
• Nonparametric Detection of Geometric Structures over Networks
• Metrizing Weak Convergence with Maximum Mean Discrepancies
• Two-sample Test using Projected Wasserstein Distance
• Unsupervised Out-of-Distribution Detection with Batch Normalization
• Distributionally Robust Optimization and Generalization in Kernel Methods
• Characterizations of non-normalized discrete probability distributions and their application in statistics
• Minimum Stein Discrepancy Estimators
• Nonparametric Empirical Bayes Estimation on Heterogeneous Data
• Sliced Kernelized Stein Discrepancy
• Stein operators, kernels and discrepancies for multivariate continuous distributions
• The reproducing Stein kernel approach for post-hoc corrected sampling
• Estimating Convergence of Markov chains with L-Lag Couplings
• Learning deep kernels for exponential family densities
• On the Exploration of Incremental Learning for Fine-grained Image Retrieval
• Learning Kernel Tests Without Data Splitting
• Kernel-based ANOVA decomposition and Shapley effects -- Application to global sensitivity analysis
• Minimum $L^q$-distance estimators for non-normalized parametric models
• Kernel Conditional Moment Test via Maximum Moment Restriction
• On Energy-Based Models with Overparametrized Shallow Neural Networks
• Blindness of score-based methods to isolated components and mixing proportions
• Two-sample Test with Kernel Projected Wasserstein Distance
• Goodness-of-fit tests on manifolds
• Kernel Stein Discrepancy Descent
• Kernelized Complete Conditional Stein Discrepancy
• Generalized Kernel Thinning
• Kernel Stein Generative Modeling
• Stein particle filtering
• Stein Variational Inference for Discrete Distributions
• Detecting Anomalous Event Sequences with Temporal Point Processes
• Statistical Depth Meets Machine Learning: Kernel Mean Embeddings and Depth in Functional Data Analysis
• A General Framework for Empirical Bayes Estimation in Discrete Linear Exponential Family
• Universal Hypothesis Testing with Kernels: Asymptotically Optimal Tests for Goodness of Fit
• A Stein Goodness-of-fit Test for Directional Distributions
• Composite Goodness-of-fit Tests with Kernels
• Interpreting diffusion score matching using normalizing flow
• Sampling with Mirrored Stein Operators
• Measure Transport with Kernel Stein Discrepancy
• Testing Goodness of Fit of Conditional Density Models with Kernels
• Estimating Density Models with Truncation Boundaries using Score Matching
• Doubly Robust Stein-Kernelized Monte Carlo Estimator: Simultaneous Bias-Variance Reduction and Supercanonical Convergence
• Bridging Explicit and Implicit Deep Generative Models via Neural Stein Estimators
• Active Slices for Sliced Stein Discrepancy
• Scalable Approximate Inference and Some Applications
• Asymptotically Optimal One- and Two-Sample Testing with Kernels
• Nonparametric Score Estimators
• High-Dimensional Multi-Task Averaging and Application to Kernel Mean Embedding
• Integrable Nonparametric Flows
• A Stein Goodness of fit Test for Exponential Random Graph Models
• Minimum Discrepancy Methods in Uncertainty Quantification
• Separation Results between Fixed-Kernel and Feature-Learning Probability Metrics
• Goodness-of-Fit Test for Mismatched Self-Exciting Processes
• Interpretable Stein Goodness-of-fit Tests on Riemannian Manifolds