GFlowCausal: Generative Flow Networks for Causal Discovery

Causal discovery aims to uncover causal structure among a set of variables. Score-based approaches mainly focus on searching for the best Directed Acyclic Graph (DAG) based on a predefined score function. However, most of them are not applicable on a large scale due to the limited searchability. Inspired by the active learning in generative flow networks, we propose a novel approach to learning a DAG from observational data called GFlowCausal. It converts the graph search problem to a generation problem, in which direct edges are added gradually. GFlowCausal aims to learn the best policy to generate high-reward DAGs by sequential actions with probabilities proportional to predefined rewards. We propose a plug-and-play module based on transitive closure to ensure efficient sampling. Theoretical analysis shows that this module could guarantee acyclicity properties effectively and the consistency between final states and fully-connected graphs. We conduct extensive experiments on both synthetic and real datasets, and results show the proposed approach to be superior and also performs well in a large-scale setting.

[1]  J. Wang,et al.  Reinforcement Causal Structure Learning on Order Graph , 2022, AAAI.

[2]  Bonaventure F. P. Dossou,et al.  Biological Sequence Design with GFlowNets , 2022, ICML.

[3]  Chris C. Emezue,et al.  Bayesian Structure Learning with Generative Flow Networks , 2022, UAI.

[4]  Aaron C. Courville,et al.  Generative Flow Networks for Discrete Probabilistic Modeling , 2022, ICML.

[5]  Chen Sun,et al.  Trajectory Balance: Improved Credit Assignment in GFlowNets , 2022, NeurIPS.

[6]  Doina Precup,et al.  Flow Network based Generative Models for Non-Iterative Diverse Candidate Generation , 2021, NeurIPS.

[7]  Zhitang Chen,et al.  Ordering-Based Causal Discovery with Reinforcement Learning , 2021, IJCAI.

[8]  Christof Seiler,et al.  Beware of the Simulated DAG! Causal Discovery Benchmarks May Be Easy to Game , 2021, NeurIPS.

[9]  Zhitang Chen,et al.  CausalVAE: Disentangled Representation Learning via Neural Structural Causal Models , 2020, 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[10]  Dennis Wei,et al.  DAGs with No Fears: A Closer Look at Continuous Optimization for Learning Bayesian Networks , 2020, NeurIPS.

[11]  Mihaela van der Schaar,et al.  CASTLE: Regularization via Auxiliary Causal Graph Discovery , 2020, NeurIPS.

[12]  Huan Liu,et al.  Causal Adversarial Network for Learning Conditional and Interventional Distributions. , 2020, 2008.11376.

[13]  Kun Zhang,et al.  On the Role of Sparsity and DAG Constraints for Learning Linear DAGs , 2020, NeurIPS.

[14]  Peter M. Aronow,et al.  The Book of Why: The New Science of Cause and Effect , 2020, Journal of the American Statistical Association.

[15]  Zhitang Chen,et al.  Causal Discovery with Reinforcement Learning , 2019, ICLR.

[16]  Tristan Deleu,et al.  Gradient-Based Neural DAG Learning , 2019, ICLR.

[17]  Mo Yu,et al.  DAG-GNN: DAG Structure Learning with Graph Neural Networks , 2019, ICML.

[18]  Eric V. Strobl.  A constraint-based algorithm for causal discovery with cycles, latent variables and selection bias , 2018, International Journal of Data Science and Analytics.

[19]  Mélanie Frappier,et al.  The Book of Why: The New Science of Cause and Effect , 2018, Science.

[20]  I. Guyon,et al.  Structural Agnostic Modeling: Adversarial Learning of Causal Graphs , 2018, J. Mach. Learn. Res..

[21]  Pradeep Ravikumar,et al.  DAGs with NO TEARS: Continuous Optimization for Structure Learning , 2018, NeurIPS.

[22]  Sergey Levine,et al.  Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor , 2018, ICML.

[23]  Bernhard Schölkopf,et al.  Elements of Causal Inference: Foundations and Learning Algorithms , 2017 .

[24]  Bernhard Schölkopf,et al.  Distinguishing Cause from Effect Using Observational Data: Methods and Benchmarks , 2014, J. Mach. Learn. Res..

[25]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[26]  Max Welling,et al.  Auto-Encoding Variational Bayes , 2013, ICLR.

[27]  Bernhard Schölkopf,et al.  Causal discovery with continuous additive noise models , 2013, J. Mach. Learn. Res..

[28]  Peter Bühlmann,et al.  CAM: Causal Additive Models, high-dimensional order search and penalized regression , 2013, ArXiv.

[29]  P. Bühlmann,et al.  Estimating High-Dimensional Directed Acyclic Graphs with the PC-Algorithm , 2005, J. Mach. Learn. Res..

[30]  Aapo Hyvärinen,et al.  A Linear Non-Gaussian Acyclic Model for Causal Discovery , 2006, J. Mach. Learn. Res..

[31]  Daphne Koller,et al.  Ordering-Based Search: A Simple and Effective Algorithm for Learning Bayesian Networks , 2005, UAI.

[32]  K. Sachs,et al.  Causal Protein-Signaling Networks Derived from Multiparameter Single-Cell Data , 2005, Science.

[33]  Paul Humphreys,et al.  Are There Algorithms That Discover Causal Structure? , 1999, Synthese.

[34]  David Maxwell Chickering,et al.  Learning Bayesian Networks: The Combination of Knowledge and Statistical Data , 1994, Machine Learning.

[35]  Dirk Husmeier,et al.  Sensitivity and specificity of inferring genetic regulatory interactions from microarray experiments with dynamic Bayesian networks , 2003, Bioinform..

[36]  Tom Burr,et al.  Causation, Prediction, and Search , 2003, Technometrics.

[37]  David Maxwell Chickering,et al.  Optimal Structure Identification With Greedy Search , 2002, J. Mach. Learn. Res..

[38]  David Maxwell Chickering,et al.  Learning Bayesian Networks is , 1994 .

[39]  Wai Lam,et al.  LEARNING BAYESIAN BELIEF NETWORKS: AN APPROACH BASED ON THE MDL PRINCIPLE , 1994, Comput. Intell..

[40]  David Heckerman,et al.  Learning Gaussian Networks , 1994, UAI.