Two optimal strategies for active learning of causal models from interventional data

Hauser, Alain; Bühlmann, Peter

doi:10.1016/j.ijar.2013.11.007

Cited by 81 publications

(135 citation statements)

References 16 publications

Supporting

Mentioning

134

Contrasting

Order By: Relevance

“…We leverage an objective Bayes procedure, whose output is a posterior distribution on the space of interventional Markov equivalence classes, to construct a sequential optimal design criterion based on posterior inclusion probabilities of as yet undirected edges computed across the whole collection of essential graphs. In this way we bypass the bottleneck of selecting up front or, at each step of a sequential procedure, a single graph which represents a major source of estimation error in active learning designs that are employed at present (Hauser and Bühlmann, 2014). We demonstrate through simulation that our criterion leads to the discovery of the true generating DAG through a small number of sequential interventions.…”

Section: Discussionmentioning

confidence: 99%

“…More recently, Hauser and Bühlmann (2014) presented an active learning strategy for causal discovery based on the characterization of interventional Markov equivalence classes of DAGs. In particular, they proposed a greedy approach that maximizes at each intervention the number of edges that can be oriented.…”

Section: Optimal Intervention Target Selectionmentioning

confidence: 99%

“…Specifically Tong and Koller (2001) dealt with discrete random variables only and did not make explicit use of the notion of equivalence classes; in contrast Masegosa and Moral (2013) relied on substantive (error-free) expert knowledge to select variables. Hauser and Bühlmann (2014) made a significant advance and proposed two methods for active learning. The first is a greedy approach using single-vertex interventions that maximizes the number of edges that can be oriented after each intervention.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Discovering Causal Structures in Bayesian Gaussian Directed Acyclic Graph Models

Castelletti

Consonni

2020

Journal of the Royal Statistical Society Series A: Statistics in Society

View full text Add to dashboard Cite

Causal directed acyclic graphs (DAGs) are naturally tailored to represent biological signalling pathways. However, a causal DAG is only identifiable up to Markov equivalence if only observational data are available. Interventional data, based on exogenous perturbations of the system, can greatly improve identifiability. Since the gain of an intervention crucially depends on the intervened variables, a natural issue is devising efficient strategies for optimal causal discovery. We present a Bayesian active learning procedure for Gaussian DAGs which requires no subjective specification on the side of the user, explicitly takes into account the uncertainty on the space of equivalence classes (through the posterior distribution) and sequentially proposes the choice of the optimal intervention variable. In simulation experiments our method, besides surpassing designs based on a random choice of intervention nodes, shows decisive improvements over currently available algorithms and is competitive with the best alternative benchmarks. An important reason behind this strong performance is that, unlike non-Bayesian algorithms, our utility function naturally incorporates graph estimation uncertainty through the posterior edge inclusion probability. We also reanalyse the Sachs data on protein signalling pathways from an active learning perspective and show that DAG identification can be achieved by using only a subset of the available intervention samples.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Optimal Intervention Target Selectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Discovering Causal Structures in Bayesian Gaussian Directed Acyclic Graph Models

Castelletti

Consonni

2020

Journal of the Royal Statistical Society Series A: Statistics in Society

View full text Add to dashboard Cite

show abstract

“…Neither do we consider the addition/integration of biological knowledge or of complimentary data sets [144] or a supervised framework [95]. Lastly, we do not cover purely interventional designs [32,54,88]. In some sense, our work is related to the work of [94] or that of [6], but we explore beyond the cause-effect pair of variables or the treatment effect.…”

Section: Data and Reconstruction Methodsmentioning

confidence: 99%

Causal Queries from Observational Data in Biological Systems via Bayesian Networks: An Empirical Study in Small Networks

White

Vignes

2018

Methods in Molecular Biology

View full text Add to dashboard Cite

Biological networks are a very convenient modelling and visualisation tool to discover knowledge from modern high-throughput genomics and postgenomics data sets. Indeed, biological entities are not isolated, but are components of complex multi-level systems. We go one step further and advocate for the consideration of causal representations of the interactions in living systems. We present the causal formalism and bring it out in the context of biological networks, when the data is observational. We also discuss its ability to decipher the causal information flow as observed in gene expression. We also illustrate our exploration by experiments on small simulated networks as well as on a real biological data set.

show abstract

“…We contribute to this line of work by deriving the first Bayesian active learning algorithm for GBNs, where the informativeness of each candidate intervention is estimated via Bayesian inference, treating the graph as a latent random variable, and the most informative intervention is chosen. In the non-Bayesian setting, Hauser et al [ 13 ], Eberhardt [ 2 ], and He and Geng [ 14 ] proposed active learning algorithms based on graph-theoretic insights, where the goal is to orient the most number of undirected edges in a Markov equivalence class with an intervention. Notably, these approaches aim only to determine the direction of edges in a given undirected graph (skeleton) estimated from observational data, and thus cannot handle errors already incorporated into the skeleton as a result of limited sample sizes and noisy observations.…”

Section: Introductionmentioning

confidence: 99%

Reconstructing Causal Biological Networks through Active Learning

2016

View full text Add to dashboard Cite

Reverse-engineering of biological networks is a central problem in systems biology. The use of intervention data, such as gene knockouts or knockdowns, is typically used for teasing apart causal relationships among genes. Under time or resource constraints, one needs to carefully choose which intervention experiments to carry out. Previous approaches for selecting most informative interventions have largely been focused on discrete Bayesian networks. However, continuous Bayesian networks are of great practical interest, especially in the study of complex biological systems and their quantitative properties. In this work, we present an efficient, information-theoretic active learning algorithm for Gaussian Bayesian networks (GBNs), which serve as important models for gene regulatory networks. In addition to providing linear-algebraic insights unique to GBNs, leading to significant runtime improvements, we demonstrate the effectiveness of our method on data simulated with GBNs and the DREAM4 network inference challenge data sets. Our method generally leads to faster recovery of underlying network structure and faster convergence to final distribution of confidence scores over candidate graph structures using the full data, in comparison to random selection of intervention experiments.

show abstract

Two optimal strategies for active learning of causal models from interventional data

Cited by 81 publications

References 16 publications

Discovering Causal Structures in Bayesian Gaussian Directed Acyclic Graph Models

Discovering Causal Structures in Bayesian Gaussian Directed Acyclic Graph Models

Causal Queries from Observational Data in Biological Systems via Bayesian Networks: An Empirical Study in Small Networks

Reconstructing Causal Biological Networks through Active Learning

Contact Info

Product

Resources

About