Kernel-based Conditional Independence Test and Application in Causal Discovery

2012/02/14 by Kun Zhang, Jonas Peters, Zhang, Kun +4 · 40 citations
Computer Science · #Bayesian Modeling and Causal Inference #FOS: Computer and information sciences #Geochemistry and Geologic Mapping #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Rough Sets and Fuzzy Logic

paper · pdf · doi:10.48550/arxiv.1202.3775

openalex publication_date 2012/02/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Conditional independence testing is an important problem, especially in Bayesian network learning and causal discovery. Due to the curse of dimensionality, testing for conditional independence of continuous variables is particularly challenging. We propose a Kernel-based Conditional Independence test (KCI-test), by constructing an appropriate test statistic and deriving its asymptotic distribution under the null hypothesis of conditional independence. The proposed method is computationally efficient and easy to implement. Experimental results show that it outperforms other methods, especially when the conditioning set is large or the sample size is not very large, in which case other methods encounter difficulties.

Citations

• On Stochastic Limit and Order Relationships

Cited by

• Ordering-based Causal Discovery via Generalized Score Matching
• Toward Scalable and Valid Conditional Independence Testing with Spectral Representations
• On the Hardness of Conditional Independence Testing In Practice
• Text Rationalization for Robust Causal Effect Estimation
• Realizing LLMs' Causal Potential Requires Science-Grounded, Novel Benchmarks
• Differentiable Constraint-Based Causal Discovery
• Local Causal Discovery for Statistically Efficient Causal Inference
• CausalVerse: Benchmarking Causal Representation Learning with Configurable High-Fidelity Simulations
• CALM: A Causal Analysis Language Model for Tabular Data in Complex Systems with Local Scores, Conditional Independence Tests, and Relation Attributes
• IRIS: An Iterative and Integrated Framework for Verifiable Causal Discovery in the Absence of Tabular Data
• DODO: Causal Structure Learning with Budgeted Interventions
• A powerful framework for differential co-expression analysis of general risk factors
• Differentiable Causal Discovery from Interventional Data
• Scalable Causal Discovery from Recursive Nonlinear Data via Truncated Basis Function Scores and Tests
• Characterization and Learning of Causal Graphs with Latent Confounders and Post-treatment Selection from Interventional Data
• Demographic influences on investment: A causal discovery study in Japan
• Causal Generative Neural Networks
• Efficient Ensemble Conditional Independence Test Framework for Causal Discovery
• Learning Neural Causal Models from Unknown Interventions
• Causal Generative Domain Adaptation Networks
• A U-statistic Approach to Hypothesis Testing for Structure Discovery in Undirected Graphical Models
• Discovery and Visualization of Nonstationary Causal Models
• A Conditional Distribution Equality Testing Framework using Deep Generative Learning
• Obtaining Causal Information by Merging Datasets with MAXENT
• On conditional parity as a notion of non-discrimination in machine learning
• Information Flow in Computational Systems
• PersonaX: Multimodal Datasets with LLM-Inferred Behavior Traits
• Conditional Independence Testing Using Exchangeable Pairs
• On a Nonparametric Notion of Residual and its Applications
• Why did the distribution change?
• A New Measure of Conditional Dependence
• Exponential Concentration of a Density Functional Estimator
• Conditional Independence Testing using Generative Adversarial Networks
• A Distribution Free Conditional Independence Test with Applications to Causal Discovery
• Quasi-Monte Carlo Feature Maps for Shift-Invariant Kernels
• Conditional Sum-Product Networks: Imposing Structure on Deep\n Probabilistic Architectures
• Causal evidence for the primordiality of colours in trans-Neptunian objects
• Uncertainty Sets for Distributionally Robust Bandits Using Structural Equation Models
• Structural Causal Models for Extremes: an Approach Based on Exponent Measures
• Predictive Independence Testing, Predictive Conditional Independence Testing, and Predictive Graphical Modelling

Related

• A Fast Kernel-based Conditional Independence test with Application to Causal Discovery
• Approximate Kernel-based Conditional Independence Tests for Fast\n Non-Parametric Causal Discovery
• A Distribution Free Conditional Independence Test with Applications to Causal Discovery
• Conditional independence testing based on a nearest-neighbor estimator of conditional mutual information
• Practical Kernel Tests of Conditional Independence