NODAGS-Flow: Nonlinear Cyclic Causal Structure Learning

Muralikrishnna G Sethuraman, Romain Lopez, Rahul Mohan, Faramarz Fekri, Tommaso Biancalani, Jan-Christian Huetter
Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, PMLR 206:6371-6387, 2023.

Abstract

Learning causal relationships between variables is a well-studied problem in statistics, with many important applications in science. However, modeling real-world systems remain challenging, as most existing algorithms assume that the underlying causal graph is acyclic. While this is a convenient framework for developing theoretical developments about causal reasoning and inference, the underlying modeling assumption is likely to be violated in real systems, because feedback loops are common (e.g., in biological systems). Although a few methods search for cyclic causal models, they usually rely on some form of linearity, which is also limiting, or lack a clear underlying probabilistic model. In this work, we propose a novel framework for learning nonlinear cyclic causal graphical models from interventional data, called NODAGS-Flow. We perform inference via direct likelihood optimization, employing techniques from residual normalizing flows for likelihood estimation. Through synthetic experiments and an application to single-cell high-content perturbation screening data, we show significant performance improvements with our approach compared to state-of-the-art methods with respect to structure recovery and predictive performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v206-sethuraman23a, title = {NODAGS-Flow: Nonlinear Cyclic Causal Structure Learning}, author = {Sethuraman, Muralikrishnna G and Lopez, Romain and Mohan, Rahul and Fekri, Faramarz and Biancalani, Tommaso and Huetter, Jan-Christian}, booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics}, pages = {6371--6387}, year = {2023}, editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem}, volume = {206}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v206/sethuraman23a/sethuraman23a.pdf}, url = {https://proceedings.mlr.press/v206/sethuraman23a.html}, abstract = {Learning causal relationships between variables is a well-studied problem in statistics, with many important applications in science. However, modeling real-world systems remain challenging, as most existing algorithms assume that the underlying causal graph is acyclic. While this is a convenient framework for developing theoretical developments about causal reasoning and inference, the underlying modeling assumption is likely to be violated in real systems, because feedback loops are common (e.g., in biological systems). Although a few methods search for cyclic causal models, they usually rely on some form of linearity, which is also limiting, or lack a clear underlying probabilistic model. In this work, we propose a novel framework for learning nonlinear cyclic causal graphical models from interventional data, called NODAGS-Flow. We perform inference via direct likelihood optimization, employing techniques from residual normalizing flows for likelihood estimation. Through synthetic experiments and an application to single-cell high-content perturbation screening data, we show significant performance improvements with our approach compared to state-of-the-art methods with respect to structure recovery and predictive performance.} }
Endnote
%0 Conference Paper %T NODAGS-Flow: Nonlinear Cyclic Causal Structure Learning %A Muralikrishnna G Sethuraman %A Romain Lopez %A Rahul Mohan %A Faramarz Fekri %A Tommaso Biancalani %A Jan-Christian Huetter %B Proceedings of The 26th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2023 %E Francisco Ruiz %E Jennifer Dy %E Jan-Willem van de Meent %F pmlr-v206-sethuraman23a %I PMLR %P 6371--6387 %U https://proceedings.mlr.press/v206/sethuraman23a.html %V 206 %X Learning causal relationships between variables is a well-studied problem in statistics, with many important applications in science. However, modeling real-world systems remain challenging, as most existing algorithms assume that the underlying causal graph is acyclic. While this is a convenient framework for developing theoretical developments about causal reasoning and inference, the underlying modeling assumption is likely to be violated in real systems, because feedback loops are common (e.g., in biological systems). Although a few methods search for cyclic causal models, they usually rely on some form of linearity, which is also limiting, or lack a clear underlying probabilistic model. In this work, we propose a novel framework for learning nonlinear cyclic causal graphical models from interventional data, called NODAGS-Flow. We perform inference via direct likelihood optimization, employing techniques from residual normalizing flows for likelihood estimation. Through synthetic experiments and an application to single-cell high-content perturbation screening data, we show significant performance improvements with our approach compared to state-of-the-art methods with respect to structure recovery and predictive performance.
APA
Sethuraman, M.G., Lopez, R., Mohan, R., Fekri, F., Biancalani, T. & Huetter, J.. (2023). NODAGS-Flow: Nonlinear Cyclic Causal Structure Learning. Proceedings of The 26th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 206:6371-6387 Available from https://proceedings.mlr.press/v206/sethuraman23a.html.

Related Material