A Constraint-Propagation Approach to Probabilistic Reasoning

By Judea Pearl. 1985

The paper demonstrates that strict adherence to probability theory does not preclude the use of concurrent, self-activated constraint-propagation mechanisms for managing uncertainty. Maintaining local records of sources-of-belief allows both predictive and diagnostic inferences to be activated simultanously and propagate harmoniously towards a stable equillibrium.

How researchers see this paper

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

Citations in the corpus

Invariant Causal Prediction for Nonlinear Models (2017, background): “aph Gif the following statement holds for all disjoint sets A, B, and Cof variables: Ais independent of B, given C, if and only if Ais d-separated (in G) from B, given C. The concept of d-separation (Pearl, 1985, 1988) is dened in Peters et al. (2017, Def. 6.1), for example. 3. Also see Heinze-Deml ”
Foundations of structural causal models with cycles and latent variables (2016, background): “6), also known as the d-separation criterion [47], is one of the most widely used.”
Joint Causal Inference from Multiple Contexts (2016, background): “One of the most common formulations of such Markov properties involves the following notion of d-separation, first proposed by Pearl (1986) in the context of DAGs, and later shown to be more generally applicable:7”
Learning Bayesian networks: approaches and issues (2011, methodology): “Pearl (1986a) was the first to use this method, while Suermondt and Cooper (1988, 1990) show the optimal cutset is NP-hard to find.”
Learning Gaussian Graphical Models with Observed or Latent FVSs (2013, methodology): “Both inference and learning are efficient for tree-structured graphs (graphs without cycles): inference can be computed exactly in linear time (with respect to the size of the graph) using belief propagation (BP) [1] while the learning problem can be solved exactly in quadratic time using the Chow-L”
Causal Inference from Graphical Models (2001, methodology): “An alternative formulation of the global, directed Markov property was given by Pearl (1986a) with a formal treatment in Verma and Pearl (1990).”
Statistical Causality from a Decision-Theoretic Perspective (2014, methodology): “We here make use of the notation of Pearl (2009) in which e.”
Causal networks: semantics and expressiveness (2013, methodology): “Third, there exist various efficient algorithms that make implicit use of these representation schemes [6] [7] [8] [9] [12] [13].”
Simulation Approaches to General Probabilistic Inference on Belief Networks (2013, methodology): “For example, there are a variety of exact algorithms for general networks, using clique j oin trees [Lauritzen and Spiegelhalter 1988], conditioning [Pearl1986b] or arc reversal [Shachter 1986].”
Fusion, Propagation, and Structuring in Belief Networks (1986, methodology): “A third method called conditioning [22] is based on our ability to change the connectivity of a network and render it singly connected by instantiating a selected group of variables.”