Distinguishing Cause from Effect Using Observational Data: Methods and Benchmarks

By Jan Peters, J. Janzing, D. Zscheischler, J. Schölkopf, Joris M. Mooij, Dominik Janzing, Jakob Zscheischler, Bernhard Schölkopf. 2014

445 citations recorded by the source corpus.

Subfields: Causal Inference & Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Causal Inference & Discovery

Citations in the corpus

Toward Causal Representation Learning (2021, background): “On the other hand, although methods for learning causal structure from observations exist [18], [37], [83], [113], [123], [139], [161], [174]–[176], [188]–[190], [229], [237], [246], [279], learning causal relations frequently requires collecting data from multiple environments or the ability to…”
Kernel Mean Embedding of Distributions: A Review and Beyonds (2016, background): “There has recently been a great deal of research and interest on causality in the machine learning community (Schölkopf et al. 2012, Guyon 2013, Mooij et al. 2016, Guyon 2014).”
Discovering Causal Signals in Images (2016, methodology): “In this manner, we generalize the homoscedastic noise assumption ubiquitous in previous literature [19].”
Towards Causal Representation Learning (2021, background): “[237, 188, 229, 113, 174, 187, 139, 17, 246, 277, 175, 123, 186, 176, 36, 82, 161], learning causal relations frequently requires collecting data from multiple environments, or the ability to perform interventions [251].”
Detecting non-causal artifacts in multivariate linear regression models (2018, neutral): “Each of these method relies on idealized assumptions that rarely hold in practice; never-theless they can be useful if the methods possess a degree of robustness regarding violation of model assumptions Mooij et al. (2016).”
Causal Generative Neural Networks (2017, methodology): “CE-Gauss contains 300 artiﬁcial cause-effect pairs as generated by [Mooij et al. , 2016], using random mixtures of Gaussians as causes, and Gaussian process priors as causal mechanisms.”
Causal Discovery from Temporally Aggregated Time Series (2017, neutral): “We conducted experiments on the Temperature Ozone data (Mooij et al., 2016) and the macroeconomic data used in (Moneta, 2008).”
The Arrow of Time in Multivariate Time Series (2016, background): “Alternatively, one can require that the VARMA process is in the so called ﬁnal equations or echelon form (L¨utkepohl, 2010, Section 12.1.2).”
Kernel‐based tests for joint independence (2016, methodology): “Given 349 measurements of the variables altitude, temperature and sunshine (the data set is taken from Mooij et al. (2016), pair0001.txt and pair0004.txt), we try to determine the correct causal structure out of 25 possible DAGs.”
Computing functions of random variables via reproducing kernel Hilbert space representations (2015, neutral): “ model E = f(C)+U with C ??U where C;E;U denote cause, effect, and residual (or “unexplained”) variable, respectively. Below we present a preliminary result on the CauseEffectPairs benchmark data set [22]. 15 10 20 30 40 50 0 0.05 0.1 0.15 0.2 multiplication sample size ( m ) error ^71 ^72 ^73 10 20”
Distinguishing Cause from Effect Using Observational Data: Methods and Benchmarks (2014, background): “Y = f(X,U1, . . . , Um), X ⊥ U , X ∼ pX(x), U ∼ pU (u1, . . . , um) (1) where f : R × Rm → R is a possibly nonlinear function (measurable with respect to the Borel sets of R × Rm and R), and pX(x) and pU (u1, . . . , um) are the joint densities of the observed cause X and latent causes U (with…”
COPHY: Counterfactual Learning of Physical Dynamics (2019, background): “In the machine learning community, causal reasoning gained mainistream attention relatively recently (Lopez-Paz et al., 2017; Lopez-Paz & Oquab, 2017; Kocaoglu et al., 2018; Rojas-Carulla et al., 2018; Mooij et al., 2016; Schölkopf et al., 2012), due to limitations of statistical learning becoming i”
Causal Reasoning and Large Language Models: Opening a New Frontier for Causality (2023, methodology): “This dataset (Mooij et al., 2016) consists of data for 108 different cause-effect pairs selected from 37 datasets from various domains, including meteorology, biology, medicine, engineering, and economics (see examples in Table 1).”