Causation, prediction, and search

By Peter Spirtes, Clark Glymour, Richard Scheines. 1993

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

Gradient-Based Neural DAG Learning (2019, background): “This assumption allows one to identify, not G itself, but the Markov equivalence class to which it belongs (Spirtes et al., 2000).”
A Causal Bayesian Networks Viewpoint on Fairness (2018, neutral): “Before doing that, we give some background on CBNs [18,39,40,42,45], assuming that all variables except A are continuous.”
Causal Inference through a Witness Protection Program (2014, methodology): “umber of covariates, there is an opportunity to use reasonably broad priors and obtain informative consequences on the values of @by a more conservative exploitation of the faithfulness assumption of Spirtes et al. (2000). 7. Conclusion Our model provides a novel compromise between point estimators ”
Learning stable and predictive structures in kinetic systems (2018, background): “Then, one fits an `1-penalized sparse linear model Tibshirani et al. [2015], Donoho [2006], Candès [2006] on these estimated derivatives ̂̇ Y (i) t` = d ∑”
Causal and anti-causal learning in pattern recognition for neuroimaging (2015, background): “ot always uniquely determine the causal structure of a given set of observed variables. In general, conditional independence tests on all subsets of observed variables may provide further information [7, 13]. An exhaustive description of the causal inference rules based on conditional independence t”
Nonparametric Independence Testing for Small Sample Sizes (2014, methodology): “ependence when there isn’t any). Independence testing is often a precursor to further analysis. Consider for instance conditional independence testing for inferring causality, say by the PC algorithm Spirtes et al. (2000), whose ﬁrst step is (unconditional) independence testing. It is also useful fo”
Causal Inference Using the Algorithmic Markov Condition (2008, methodology): “Apart from that, the most important difference between human causal learning and the inference rules in [ 2 , 1] is that the former is also about causal relations among single objects and does not necessarily require sampling.”
CAM: Causal Additive Models, high-dimensional order search and penalized regression (2013, methodology): “ponding parameters) are greedy equivalent search for computing the BIC-regularized maximum likelihood estimator (Chickering, 2002) or the PC-algorithm using multiple conditional independence testing (Spirtes et al., 2000). However, for the latter, the constraint of additive structural equations cann”
Identifiability of Gaussian structural equation models with equal error variances (2012, methodology): “1 Existing methods We compare our method against the PC-algorithm [Spirtes et al., 2000] and greedy equivalence search [Chickering, 2002].”
Identifying confounders using additive noise models (2009, methodology): “Since the pioneering work on causal inference methods (described for example in [Pearl, 2000] and [Spirtes et al., 1993]), much work has been done under the assumption that all relevant variables have been observed.”
Disentangling Factors of Variation Using Few Labels (2019, background): “In particular, we hope to learn a disentangled representation with as little supervision as possible so that all the available labels can be used to learn downstream tasks [2, 57, 49, 48, 58].”
Characterization and greedy learning of interventional Markov equivalence classes of directed acyclic graphs (2011, methodology): “Several algorithms for estimating essential graphs from observational data exist, such as the PC algorithm (Spirtes et al., 2000) or the Greedy Equivalence Search (GES) algorithm (Meek, 1997; Chickering, 2002b); a more complete overview is given in Brown et al. (2005) and Murphy (2001).”
Causal interpretation of stochastic differential equations (2013, methodology): “ions are identiﬁable from the generator. In the remainder of this section, we discuss the content of Theorem 5.3. First, recall that a main theme of the DAG-based framework for causal inference as in [35, 26] is to identify conditions for when postintervention distributions are identiﬁable from the ”
Strong completeness and faithfulness in Bayesian networks (1995, methodology): “…same set of variables V such that there exists a factorization of P(V) such that P(V) = IJ P(AJparents(A)) AEV where P(AJparents(A)) is a conditional probability density and parents(A) is the set of parents of vertex A in graph G. Linear Bayesian networks are described in Spirtes et al. (1993) .”