Local Causal and Markov Blanket Induction for Causal Discovery and Feature Selection for Classification Part I: Algorithms and Empirical Evaluation

By Constantin F. Aliferis, Alexander R. Statnikov, Ioannis Tsamardinos, Subramani Mani, Xenofon D. Koutsoukos. 2010

611 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

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 Statistics and Probability

Citations in the corpus

Causal Modeling-Based Discrimination Discovery and Removal: Criteria, Bounds, and Algorithms (2019, background): “Many researches have been devoted to improving the performance of network construction [30], [32], [33] and probabilistic inference in causal graphs [34], [35].”
A Causal Framework for Discovering and Removing Direct and Indirect Discrimination (2016, background): “Many researches have been devoted to improving the performance of network construction (Kalisch and B¨uhlmann 2007; Tsamardinos et al. 2003; Aliferis et al. 2010) and probabilistic inference in causal networks (Heckerman and Breese 1994; 1996).”
New methods for separating causes from effects in genomics data (2012, methodology): “A rigorous approach would involve first using correct causal discovery methods (e.g., GLL/ LGL [6,7], MMHC [19], PC/FCI [1], IC/IC* [2], etc.) to identify unoriented edges that denote the existence of unconfounded causal relations and then applying causal orientation techniques to orient unoriented…”
Advances in variable selection methods I: Causal selection methods versus stepwise regression and principal component analysis on data of known and unknown functional relationships (2012, methodology): “For details readers may refer to Fu and Desmarais (2010) and Aliferis et al. (2010).”
Network analysis methods for studying microbial communities: A mini review (2021, methodology): “FlashWeave [54] mitigates this issue by using a modified version of the semi-interleaved HITON-PC algorithm [55], a causal inference algorithm which infers for each taxon its Markov blanket.”
Estimating the effect of joint interventions from observational data in sparse high-dimensional settings (2014, background): “For very large graphs, one could even go a step further and only learn the Markov blankets of the intervention nodes [1, 3, 28] and extract possible parent sets from there.”
Feature Selection: A literature Review (2014, methodology): “Related to the high-dimensional dataset (in both samples and attributes), the performance of feature selection methods are studied for the multiple-class problem [90, 97, 98, 99].”
A review of feature selection methods on synthetic data (2013, background): “, a high number of features versus low number of samples; most of these studies also tackle with the multiple class problems [24,35–38].”
A survey of Bayesian Network structure learning (2021, background): “Aliferis et al. (2010) defined a sound generic framework for learning 𝑷𝑪(𝑇) into which these three specific algorithms fit, and which consists of: • a strategy for inclusion of a node 𝑋 in 𝑷𝑪(𝑇), heuristically prioritised, for instance, based on the strength of association between 𝑋 and…”
Learning Sparse Nonparametric DAGs (2019, result): “Comparisons with other score-based methods (KGV score (Bach and Jordan, 2003), Spearman correlation (Sokolova et al., 2014)) and constraint-based methods (PC (Spirtes et al., 2000), MM-MB (Aliferis et al., 2010)) can be found in previous work (Huang et al., 2018), hence are omitted.”
Generalized Score Functions for Causal Discovery (2018, methodology): “Since the search procedure may affect the results, we applied the widely-used PC search [28] and state-of-theart search procedures, including the semi-interleaved HITON-MB search with symmetry correction [1] and the max-min Markov blanket (MM-MB) search with symmetry correction [1].”
Towards Out-Of-Distribution Generalization: A Survey (2021, methodology): “Markov boundary discovery is generally challenging because traditional methods [7, 8] are always based on the conditional independence test, which is a particularly difficult hypothesis to test for [231].”
Bayesian Networks for Variable Groups (2015, methodology): “Although the problem is NP-hard (Chickering, 1996), there exist plenty of exact algorithms (Jaakkola et al., 2010; Silander and Myllymäki, 2006) as well as theoretically sound heuristics (Aliferis et al., 2010; Chickering, 2002).”
Data Based Identification and Prediction of Nonlinear and Complex Dynamical Systems (2016, methodology): “, simulated annealing, max-min parent and children algorithm [348], Markov blanket algorithm [349], Markov boundary induction algorithm [350].”