Varieties of Causal Intervention

By Kevin B. Korb, Lucas R. Hope, Ann E. Nicholson, Karl Axnick. 2004

121 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Philosophy

How researchers see this paper

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

In Artificial Intelligence

In Philosophy

Citations in the corpus

Differentiable Causal Discovery from Interventional Data (2020, background): “A specific case that is often considered is (stochastic) perfect interventions (or hard, structural) [10, 48, 26] where p j (xj |xπG j ) = p (k) j (xj) for all j ∈ Ik, thus removing the dependencies with their parents (see Figure 1).”
Woulda, Coulda, Shoulda: Counterfactually-Guided Policy Search (2018, background): “This also includes what is known as stochastic interventions or mechanism changes (Korb et al., 2004) which generalize atomic interventions (Pearl, 2009).”
Few-Shot and Continual Learning with Attentive Independent Mechanisms (2021, background): “Such intervention is analogous to stochastic intervention [25] and dropout [46] which adds stochasticity to the training of AIM, preventing the locking down to a few mechanisms that are attended to upon initialization.”
Characterization and greedy learning of interventional Markov equivalence classes of directed acyclic graphs (2011, neutral): “We consider stochastic interventions (Korb et al., 2004) modeling the eﬀect of setting or forcing one or several random variables X I , where I ⊂ [ p ] is called the intervention target , to the value of independent random variables U I , called intervention variables .”
Anchor regression: Heterogeneous data meet causality (2018, background): “ugh the system. Shift interventions can be seen as a special case of a \parametric", \imperfect" or \dependent" intervention or a \mechanism change" [Eberhardt and Scheines, 2007, Korb et al., 2004, Tian and Pearl, 2001]. The new interventional joint distribution is denoted by (X”
Exact Bayesian structure learning from uncertain interventions (2007, background): “One possible relaxation of this model is to assume that interventions are “stochastic”, meaning that they induce a distribution over states rather than a specific state [ KHNA04 ].”
Causal Discovery and Forecasting in Nonstationary Environments with State-Space Models (2019, background): “Nonstationarity may result from a change in the underlying mechanisms, which is related to soft intervention (Korb et al., 2004) in the sense that both result in probability distribution changes, while non-stationarity can be seen as a consequence of soft interventions done by nature.”
Variable Importance and Prediction Methods for Longitudinal Problems with Missing Variables (2015, unknown): “This type of intervention has been previously discussed in the literature [23], and belongs to a wider class of interventions known as stochastic interventions [24 – 26].”
Incremental causal effects (2019, methodology): “In models based on structure equations, incremental interventions can be seen as a special case of “parametric” or “imperfect” interventions (Eberhardt and Scheines, 2007; Korb et al., 2004; Tian and Pearl, 2001).”
Algorithmic Recourse: from Counterfactual Explanations to Interventions (2020, background): “pealing and powerful, structural interventions are in many ways the simplest type of interventions, and their “simplicity comes at a price: foregoing the possibility of modeling many situations realisitically” [8, 22].”