Towards Robust Relational Causal Discovery

We consider the problem of learning causal relationships from relational data. Existing approaches rely on queries to a relational conditional independence (RCI) oracle to establish and orient causal relations in such a setting. In practice, queries to a RCI oracle have to be replaced by reliable tests for RCI against available data. Relational data present several unique challenges in testing for RCI. We study the conditions under which traditional iid-based conditional independence (CI) tests yield reliable answers to RCI queries against relational data. We show how to conduct CI tests against relational data to robustly recover the underlying relational causal structure. Results of our experiments demonstrate the effectiveness of our proposed approach.

[1]  Bernhard Schölkopf,et al.  A Kernel Method for the Two-Sample-Problem , 2006, NIPS.

[2]  Thomas S. Richardson,et al.  Causal Inference in the Presence of Latent Variables and Selection Bias , 1995, UAI.

[3]  Bernhard Schölkopf,et al.  Measuring Statistical Dependence with Hilbert-Schmidt Norms , 2005, ALT.

[4]  Alexander J. Smola,et al.  Gaussian Processes for Independence Tests with Non-iid Data in Causal Inference , 2015, ACM Trans. Intell. Syst. Technol..

[5]  P. Spirtes,et al.  Causation, prediction, and search , 1993 .

[6]  Katerina Marazopoulou,et al.  Inferring Causal Direction from Relational Data , 2016, UAI.

[7]  Joaquín Abellán,et al.  Some Variations on the PC Algorithm , 2006, Probabilistic Graphical Models.

[8]  Daphne Koller,et al.  Probabilistic Relational Models , 1999, ILP.

[9]  Vasant Honavar,et al.  Towards Conditional Independence Test for Relational Data , 2017, UAI.

[10]  Joris M. Mooij,et al.  Ancestral Causal Inference , 2016, NIPS.

[11]  Vasant Honavar,et al.  Self-Discrepancy Conditional Independence Test , 2017, UAI.

[12]  Diego Colombo,et al.  Order-independent constraint-based causal structure learning , 2012, J. Mach. Learn. Res..

[13]  D. Rubin.  Estimating causal effects of treatments in randomized and nonrandomized studies. , 1974 .

[14]  Frederick Eberhardt,et al.  Discovering Cyclic Causal Models with Latent Variables: A General SAT-Based Procedure , 2013, UAI.

[15]  Brian J. Taylor,et al.  Learning Causal Models of Relational Domains , 2010, AAAI.

[16]  Ioannis Tsamardinos,et al.  Constraint-based causal discovery from multiple interventions over overlapping variable sets , 2014, J. Mach. Learn. Res..

[17]  Marc Maier,et al.  Causal Discovery for Relational Domains: Representation, Reasoning, and Learning , 2014 .

[18]  David Haussler,et al.  Convolution kernels on discrete structures , 1999 .

[19]  Dimitris Margaritis,et al.  Improving the Reliability of Causal Discovery from Small Data Sets Using Argumentation , 2009, J. Mach. Learn. Res..

[20]  Jiji Zhang,et al.  Adjacency-Faithfulness and Conservative Causal Inference , 2006, UAI.

[21]  Vasant Honavar,et al.  On Learning Causal Models from Relational Data , 2016, AAAI.

[22]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[23]  Vasant Honavar,et al.  A Characterization of Markov Equivalence Classes of Relational Causal Models under Path Semantics , 2016, UAI.

[24]  Foster J. Provost,et al.  Distribution-based aggregation for relational learning with identifier attributes , 2006, Machine Learning.

[25]  Katerina Marazopoulou,et al.  A Sound and Complete Algorithm for Learning Causal Models from Relational Data , 2013, UAI.

[26]  Judea Pearl,et al.  Equivalence and Synthesis of Causal Models , 1990, UAI.

[27]  A. Cano,et al.  A Score Based Ranking of the Edges for the PC Algorithm , 2008 .

[28]  Marek J. Druzdzel,et al.  A Hybrid Anytime Algorithm for the Construction of Causal Models From Sparse Data , 1999, UAI.