Belief polarization is said to occur when two people respond to the same evidence by updating their beliefs in opposite directions. This response is considered to be “irrational” because it involves contrary updating, a form of belief updating that appears to violate normatively optimal responding, as for example dictated by Bayes' theorem. In light of much evidence that people are capable of normatively optimal behavior, belief polarization presents a puzzling exception. We show that Bayesian networks, or Bayes nets, can (...) simulate rational belief updating. When fit to experimental data, Bayes nets can help identify the factors that contribute to polarization. We present a study into belief updating concerning the reality of climate change in response to information about the scientific consensus on anthropogenic global warming. The study used representative samples of Australian and U.S. participants. Among Australians, consensus information partially neutralized the influence of worldview, with free-market supporters showing a greater increase in acceptance of human-caused global warming relative to free-market opponents. In contrast, while consensus information overall had a positive effect on perceived consensus among U.S. participants, there was a reduction in perceived consensus and acceptance of human-caused global warming for strong supporters of unregulated free markets. Fitting a Bayes net model to the data indicated that under a Bayesian framework, free-market support is a significant driver of beliefs about climate change and trust in climate scientists. Further, active distrust of climate scientists among a small number of U.S. conservatives drives contrary updating in response to consensus information among this particular group. (shrink)

In a criminal trial, evidence is used to draw conclusions about what happened concerning a supposed crime. Traditionally, the three main approaches to modeling reasoning with evidence are argumentative, narrative and probabilistic approaches. Integrating these three approaches could arguably enhance the communication between an expert and a judge or jury. In previous work, techniques were proposed to represent narratives in a Bayesian network and to use narratives as a basis for systematizing the construction of a Bayesian (...) network for a legal case. In this paper, these techniques are combined to form a design method for constructing a Bayesian network based on narratives. This design method is evaluated by means of an extensive case study concerning the notorious Dutch case of the Anjum murders. (shrink)

The forensic two-trace problem is a perplexing inference problem introduced by Evett (J Forensic Sci Soc 27:375–381, 1987). Different possible ways of wording the competing pair of propositions (i.e., one proposition advanced by the prosecution and one proposition advanced by the defence) led to different quantifications of the value of the evidence (Meester and Sjerps in Biometrics 59:727–732, 2003). Here, we re-examine this scenario with the aim of clarifying the interrelationships that exist between the different solutions, and in this way, (...) produce a global vision of the problem. We propose to investigate the different expressions for evaluating the value of the evidence by using a graphical approach, i.e. Bayesian networks, to model the rationale behind each of the proposed solutions and the assumptions made on the unknown parameters in this problem. (shrink)

This article considers the extent to which Bayesian networks with imprecise probabilities, which are used in statistics and computer science for predictive purposes, can be used to represent causal structure. It is argued that the adequacy conditions for causal representation in the precise context—the Causal Markov Condition and Minimality—do not readily translate into the imprecise context. Crucial to this argument is the fact that the independence relation between random variables can be understood in several different ways when the joint (...) probability distribution over those variables is imprecise, none of which provides a compelling basis for the causal interpretation of imprecise Bayes nets. I conclude that there are serious limits to the use of imprecise Bayesian networks to represent causal structure. (shrink)

Chronic myeloid leukemia (CML) is a cancer of the hematopoietic system initiated by a single genetic mutation which results in the oncogenic fusion protein Bcr-Abl. Untreated, patients pass through different phases of the disease beginning with the rather asymptomatic chronic phase and ultimately culminating into blast crisis, an acute leukemia resembling phase with a very high mortality. Although many processes underlying the chronic phase are well understood, the exact mechanisms of disease progression to blast crisis are not yet revealed. In (...) this paper we develop a mathematical model of CML based on causal Bayesian networks in order to study possible disease progression mechanisms. Our results indicate that an increase of Bcr-Abl levels alone is not sufficient to explain the phenotype of blast crisis and that secondary changes such as additional mutations are necessary to explain disease progression and the poor therapy response of patients in blast crisis. (shrink)

A Bayesian network (BN) is a graphical model of uncertainty that is especially well suited to legal arguments. It enables us to visualize and model dependencies between different hypotheses and pieces of evidence and to calculate the revised probability beliefs about all uncertain factors when any piece of new evidence is presented. Although BNs have been widely discussed and recently used in the context of legal arguments, there is no systematic, repeatable method for modeling legal arguments as (...) BNs. Hence, where BNs have been used in the legal context, they are presented as completed pieces of work, with no insights into the reasoning and working that must have gone into their construction. This means the process of building BNs for legal arguments is ad hoc, with little possibility for learning and process improvement. This article directly addresses this problem by describing a method for building useful legal arguments in a consistent and repeatable way. The method complements and extends recent work by Hepler, Dawid, and Leucari (2007) on object-oriented BNs for complex legal arguments and is based on the recognition that such arguments can be built up from a small number of basic causal structures (referred to as idioms). We present a number of examples that demonstrate the practicality and usefulness of the method. (shrink)

Discrimination in decision making is prohibited on many attributes, but often present in historical decisions. Use of such discriminatory historical decision making as training data can perpetuate discrimination, even if the protected attributes are not directly present in the data. This work focuses on discovering discrimination in instances and preventing discrimination in classification. First, we propose a discrimination discovery method based on modeling the probability distribution of a class using Bayesian networks. This measures the effect of a protected (...) attribute in a subset of the dataset using the estimated probability distribution. Second, we propose a classification method that corrects for the discovered discrimination without using protected attributes in the decision process. We evaluate the discrimination discovery and discrimination prevention approaches on two different datasets. The empirical results show that a substantial amount of discrimination identified in instances is prevented in future decisions. (shrink)

The appeal to expert opinion is an argument form that uses the verdict of an expert to support a position or hypothesis. A previous scheme-based treatment of the argument form is formalized within a Bayesian network that is able to capture the critical aspects of the argument form, including the central considerations of the expert's expertise and trustworthiness. We propose this as an appropriate normative framework for the argument form, enabling the development and testing of quantitative predictions as (...) to how people evaluate this argument, suggesting that such an approach might be beneficial to argumentation research generally. We subsequently present two experiments as an example of the potential for future research in this vein, demonstrating that participants' quantitative ratings of the convincingness of a proposition that has been supported with an appeal to expert opinion were broadly consistent with the predictions of the Bayesian model. (shrink)

This paper is concerned with the multivariate stochastic volatility modeling of the stock market. We investigate a DGC-t-MSV model to find the historical volatility spillovers between nine markets, including S&P, Nasdaq, SSE, SZSE, HSI, FTSE, CAC, DAX, and Nikkei indices. We use the Bayesian network to analyze the spreading of herd behavior between nine markets. The main results are as follows: the DGC-t-MSV model we considered is a useful way to estimate the parameter and fit the data (...) well in the stock market; our computational analysis shows that the S&P and Nasdaq have higher volatility spillovers to the Shanghai and Shenzhen stock markets; the results also show that there is a strong correlation between stock markets in the same region. (shrink)

Two long-standing arguments in cognitive science invoke the assumption that holistic inference is computationally infeasible. The first is Fodor’s skeptical argument toward computational modeling of ordinary inductive reasoning. The second advocates modular computational mechanisms of the kind posited by Cosmides, Tooby and Sperber. Based on advances in machine learning related to Bayes nets, as well as investigations into the structure of scientific and ordinary information, I maintain neither argument establishes its architectural conclusion. Similar considerations also undermine Fodor’s decades-long diagnosis (...) of artificial intelligence research as confounded by an inability to circumscribe the amount of information relevant to inferential processes. This diagnosis is particularly inapposite with respect to Bayes nets, since one of their strengths as machine learning systems has been their capacity to reason probabilistically about large data sets whose size overwhelms the capacities of individual human reasoners. A general moral follows from these criticisms: Insights into artificial and human cognitive systems are likely to be cultivated by focusing greater attention on the structure and density of connections among items of information that are available to them. (shrink)

In the current paper, we propose a latent interdependence approach to modeling psychometric data in social networks. The idea of latent interdependence is adopted from social relations models, which formulate a mutual-rating process by both dyad members’ characteristics. Under the framework of the latent interdependence approach, we introduce two psychometric models: The first model includes the main effects of both rating-sender and rating-receiver, and the second model includes a latent distance effect to assess the influence from the dissimilarity between (...) the latent characteristics of both sides. The latent distance effect is quantified by the Euclidean distance between both sides’ trait scores. Both models use Bayesian estimation via Markov chain Monte Carlo. How accurately model parameters were estimated was evaluated in a simulation study. Parameter recovery results showed that all parameters were accurately recovered under most of the conditions investigated. As expected, the accuracy of model estimation was significantly improved as network size grew. Also, through analyzing empirical data, we showed how to use the estimates of model parameters to predict the latent weight of connections among group members and rebuild either a univariate or multivariate network at a latent trait level. Finally, we discuss issues regarding model comparison and offer suggestions for future studies. (shrink)

The exponential growth of social data both in volume and complexity has increasingly exposed many of the shortcomings of the conventional frequentist approach to statistics. The scientific community has called for careful usage of the approach and its inference. Meanwhile, the alternative method, Bayesian statistics, still faces considerable barriers toward a more widespread application. The bayesvl R package is an open program, designed for implementing Bayesian modeling and analysis using the Stan language’s no-U-turn (NUTS) sampler. The package (...) combines the ability to construct Bayesian network models using directed acyclic graphs (DAGs), the Markov chain Monte Carlo (MCMC) simulation technique, and the graphic capability of the ggplot2 package. As a result, it can improve the user experience and intuitive understanding when constructing and analyzing Bayesian network models. A case example is offered to illustrate the usefulness of the package for Big Data analytics and cognitive computing. (shrink)

We comment on the contributions of Dahlman and of Fenton et al., who both suggested a Bayesian approach to analyze the Simonshaven case. We argue that analyzing a full case with a Bayesian approach is not feasible, and that there are serious problems with assigning actual numbers to probabilities and priors. We also discuss the nature of Bayesian thinking in court, and the nature and interpretation of the likelihood ratio. In particular, we discuss what it could mean (...) that a likelihood ratio is in some sense uncertain. (shrink)

Causal models provide a framework for precisely representing complex causal structures, where specific models can be used to efficiently predict, infer, and explain the world. At the same time, we often do not know the full causal structure a priori and so must learn it from data using a causal model search algorithm. This chapter provides a general overview of causal models and their uses, with a particular focus on causal graphical models (the most commonly used causal modeling framework) (...) and methods for learning such models from observational and experimental data. The chapter concludes with two examples of productive causal search and modeling in experimental philosophy. (shrink)

Models are a principle instrument of modern science. They are built, applied, tested, compared, revised and interpreted in an expansive scientific literature. Throughout this paper, I will argue that models are also a valuable tool for the philosopher of science. In particular, I will discuss how the methodology of Bayesian Networks can elucidate two central problems in the philosophy of science. The first thesis I will explore is the variety-of-evidence thesis, which argues that the more varied the supporting evidence, (...) the greater the degree of confirmation for a given hypothesis. However, when investigated using Bayesian methodology, this thesis turns out not to be sacrosanct. In fact, under certain conditions, a hypothesis receives more confirmation from evidence that is obtained from one rather than more instruments, and from evidence that confirms one rather than more testable consequences of the hypothesis. The second challenge that I will investigate is scientific theory change. This application highlights a different virtue of modeling methodology. In particular, I will argue that Bayesian modeling illustrates how two seemingly unrelated aspects of theory change, namely the (Kuhnian) stability of (normal) science and the ability of anomalies to over turn that stability and lead to theory change, are in fact united by a single underlying principle, in this case, coherence. In the end, I will argue that these two examples bring out some metatheoretical reflections regarding the following questions: What are the differences between modeling in science and modeling in philosophy? What is the scope of the modeling method in philosophy? And what does this imply for our understanding of Bayesianism? (shrink)

This research employs the Bayesian network modeling approach, and the Markov chain Monte Carlo technique, to learn about the role of lies and violence in teachings of major religions, using a unique dataset extracted from long-standing Vietnamese folktales. The results indicate that, although lying and violent acts augur negative consequences for those who commit them, their associations with core religious values diverge in the final outcome for the folktale characters. Lying that serves a religious mission of either (...) Confucianism or Taoism (but not Buddhism) brings a positive outcome to a character (βT_and_Lie_O= 2.23; βC_and_Lie_O= 1.47; βT_and_Lie_O= 2.23). A violent act committed to serving Buddhist missions results in a happy ending for the committer (βB_and_Viol_O= 2.55). What is highlighted here is a glaring double standard in the interpretation and practice of the three teachings: the very virtuous outcomes being preached, whether that be compassion and meditation in Buddhism, societal order in Confucianism, or natural harmony in Taoism, appear to accommodate two universal vices—violence in Buddhism and lying in the latter two. These findings contribute to a host of studies aimed at making sense of contradictory human behaviors, adding the role of religious teachings in addition to cognition in belief maintenance and motivated reasoning in discounting counterargument. (shrink)

Michael G. Titelbaum presents a new Bayesian framework for modeling rational degrees of belief—the first of its kind to represent rational requirements on agents who undergo certainty loss.

We appeal to the theory of Bayesian Networks to model different strategies for obtaining confirmation for a hypothesis from experimental test results provided by less than fully reliable instruments. In particular, we consider (i) repeated measurements of a single test consequence of the hypothesis, (ii) measurements of multiple test consequences of the hypothesis, (iii) theoretical support for the reliability of the instrument, and (iv) calibration procedures. We evaluate these strategies on their relative merits under idealized conditions and show some (...) surprising repercussions on the variety-of-evidence thesis and the Duhem-Quine thesis. (shrink)

Zuidema et al. illustrate how empirical AGL studies can benefit from computational models and techniques. Computational models can help clarifying theories, and thus in delineating research questions, but also in facilitating experimental design, stimulus generation, and data analysis. The authors show, with a series of examples, how computational modeling can be integrated with empirical AGL approaches, and how model selection techniques can indicate the most likely model to explain experimental outcomes.

There is a long philosophical tradition of addressing questions in philosophy of science and epistemology by means of the tools of Bayesian probability theory (see Earman (1992) and Howson and Urbach (1993)). In the late '70s, an axiomatic approach to conditional independence was developed within a Bayesian framework. This approach in conjunction with developments in graph theory are the two pillars of the theory of Bayesian Networks, which is a theory of probabilistic reasoning in artificial intelligence. The (...) theory has been very successful over the last two decades and has found a wide array of applications ranging from medical diagnosis to safety systems for hazardous industries. (shrink)

In a criminal trial, a judge or jury needs to reason about what happened based on the available evidence, often including statistical evidence. While a probabilistic approach is suitable for analysing the statistical evidence, a judge or jury may be more inclined to use a narrative or argumentative approach when considering the case as a whole. In this paper we propose a combination of two approaches, combining Bayesian networks with scenarios. Whereas a Bayesian network is a popular (...) tool for analysing parts of a case, constructing and understanding a network for an entire case is not straightforward. We propose an explanation method for understanding a Bayesian network in terms of scenarios. This method builds on a previously proposed construction method, which we slightly adapt with the use of scenario schemes for the purpose of explaining. The resulting structure is explained in terms of scenarios, scenario quality and evidential support. A probabilistic interpretation of scenario quality is provided using the concept of scenario schemes. Finally, the method is evaluated by means of a case study. (shrink)

Lexical ambiguity can be syntactic if it involves more than one grammatical category for a single word, or semantic if more than one meaning can be associated with a word. In this article we discuss the application of a Bayesian‐network model in the resolution of lexical ambiguities of both types. The network we propose comprises a parsing subnetwork, which can be constructed automatically for any context‐free grammar, and a subnetwork for semantic analysis, which, in the spirit of (...) Fillmore's (1968) case grammars, seeks to fulfill the required cases of all candidates for verb of the sentence. Solving for the highest joint probability of the variables conditioned upon the evidences to the network yields the most likely candidate with its meaning, along with its cases and respective meanings. This is achieved by fixing the values of all evidence nodes concurrently, and then performing a stochastic simulation in which the remaining nodes are updated probabilistically with a high degree of parallelism. The process of disambiguation is directed neither by the syntax nor the semantics, but rather by the interrelation between the two subnetworks. The use of a Bayesian‐network model allows us to express this interrelation between the two subnetworks and among their constituents in a rather direct and rigorous way that, in connection with the convergence properties of the stochastic simulation, reveals a very robust model. (shrink)

By identifying and pursuing analogies between causal and logical influence I show how the Bayesian network formalism can be applied to reasoning about logical deductions.

Proponents of various causal exclusion arguments claim that for any given event, there is often a unique level of granularity at which that event is caused. Against these causal exclusion arguments, causal ecumenists argue that the same event or phenomenon can be caused at multiple levels of granularity. This paper argues that the Bayesian network approach to representing the causal structure of target systems is consistent with causal ecumenism. Given the ubiquity of Bayesian networks as a tool (...) for representing causal structure in both philosophy of science and science itself, this result speaks in favor of the ecumenical view, and against rival exclusionary accounts. Gebharter’s (Philos Phenomenol Res 95(2):353–375, 2017) argument that the Bayes nets formalism is consistent with causal exclusion is considered and rebutted. (shrink)

The coherentist theory of justification provides a response to the sceptical challenge: even though the independent processes by which we gather information about the world may be of dubious quality, the internal coherence of the information provides the justification for our empirical beliefs. This central canon of the coherence theory of justification is tested within the framework of Bayesian networks, which is a theory of probabilistic reasoning in artificial intelligence. We interpret the independence of the information gathering processes (IGPs) (...) in terms of conditional independences, construct a minimal sufficient condition for a coherence ranking of information sets and assess whether the confidence boost that results from receiving information through independent IGPs is indeed a positive function of the coherence of the information set. There are multiple interpretations of what constitute IGPs of dubious quality. Do we know our IGPs to be no better than randomization processes? Or, do we know them to be better than randomization processes but not quite fully reliable, and if so, what is the nature of this lack of full reliability? Or, do we not know whether they are fully reliable or not? Within the latter interpretation, does learning something about the quality of some IGPs teach us anything about the quality of the other IGPs? The Bayesian-network models demonstrate that the success of the coherentist canon is contingent on what interpretation one endorses of the claim that our IGPs are of dubious quality. (shrink)

We construct a probabilistic coherence measure for information sets which determines a partial coherence ordering. This measure is applied in constructing a criterion for expanding our beliefs in the face of new information. A number of idealizations are being made which can be relaxed by an appeal to Bayesian Networks.

Through their transcript products genes regulate the rates at which an immense variety of transcripts and subsequent proteins occur. Understanding the mechanisms that determine which genes are expressed, and when they are expressed, is one of the keys to genetic manipulation for many purposes, including the development of new treatments for disease. Viewing each gene in a genome as a distinct variable that is either on or off, or more realistically as a continuous variable, the values of some of these (...) variables influence the values of others through the regulatory proteins they express, including, of course, the possibility that the rate of expression of a gene at one time may, in various circumstances, influence the rate of expression of that same gene at a later time. If we imagine an arrow drawn from each gene expression variable at a given time to a gene variable whose expression it influences a short while after, the result is a network, technically a directed acyclic graph. For example, the DAG in Figure 1 is a representation of a system in which the expression level of gene G1 at time 1 ) causes the expression level of G2, which in turn causes the expression level of G3. The arrows in Figure 1 which do not have a variable at their tails are “error terms” which represent all of the causes of a variable other than the ones explicitly represented in the DAG. The DAG describes more than associations—it describes causal connections among gene expression rates. A shock to a cell—by mutation, heating, chemical treatment, etc. may alter the DAG describing the relations among gene expressions, for example by activating a gene that was otherwise not expressed, producing a cascade of new expression effects. Although “knockout” experiments can reveal some of the underlying causal network of gene expression levels, unless guided by information from other sources, such experiments are limited in how much of the network structure they can reveal, due to the sheer number of possible combinations of experimental manipulations of genes necessary to reveal the complete causal network. Recent developments have made it possible to compare quantitatively the expression of tens of thousands of genes in cells from different sources in a single experiment, and to trace gene expression over time in thousands of genes simultaneously. cDNA microarrays are already producing extensive data, much of it available on the web. Thus there are calls for analytic software that can be applied to microarray and other data to help infer regulatory networks. In this paper we will review current techniques that are available for searching for the causal relations between variables, describe algorithmic and data gathering obstacles to applying these techniques to gene expression levels, and describe the prospects for overcoming these obstacles. (shrink)

This paper shows how an efficient and parallel algorithm for inference in Bayesian Networks (BNs) can be built and implemented combining sparse matrix factorization methods with variable elimination algorithms for BNs. This entails a complete separation between a first symbolic phase, and a second numerical phase.

Embryonic development and adult tissue homeostasis are controlled through activation of intracellular signal transduction pathways by extracellular growth factors. In the past, signal transduction has largely been regarded as a linear process. However, more recent data from large‐scale and high‐throughput experiments indicate that there is extensive cross‐talk between individual signaling cascades leading to the notion of a signaling network. The behavior of such complex networks cannot be predicted by simple intuitive approaches but requires sophisticated models and computational simulations. The (...) purpose of such models is to generate experimentally testable hypotheses and to find explanations for unexpected experimental results. Here, we discuss the need for, and the future impact of, mathematical models for exploring signal transduction in different biological contexts such as for example development. BioEssays 30:1110–1125, 2008. © 2008 Wiley Periodicals, Inc. (shrink)

Bayesian networks may either be treated purely formally or be given an interpretation. I argue that current foundations are problematic, and put forward new foundations which involve aspects of both the interpreted and the formal approaches.

Because ports are considered to be the heart of the maritime transportation system, thereby assessing port performance is necessary for a nation’s development and economic success. This study proposes a novel metric, namely, “_port performance index (PPI)_”, to determine the overall performance and utilization of inland waterway ports based on six criteria,_ port facility, port availability, port economics, port service, port connectivity, and port environment_. Unlike existing literature, which mainly ranks ports based on quantitative factors, this study utilizes a (...) class='Hi'>Bayesian Network (BN) model that focuses on both quantitative and qualitative factors to rank a port. The assessment of inland waterway port performance is further analyzed based on different advanced techniques such as sensitivity analysis and belief propagation. Insights drawn from the study show that all the six criteria are necessary to predict PPI. The study also showed that port service has the highest impact while port economics has the lowest impact among the six criteria on PPI for inland waterway ports. (shrink)

This research employs the Bayesian network modeling approach, and the Markov chain Monte Carlo technique, to learn about the role of lies and violence in teachings of major religions, using a unique dataset extracted from long-standing Vietnamese folktales. The results indicate that, although lying and violent acts augur negative consequences for those who commit them, their associations with core religious values diverge in the outcome for the folktale characters. Lying that serves a religious mission of either Confucianism (...) or Taoism (but not Buddhism) brings a positive outcome to a character. A violent act committed to serving Buddhist mission results in a happy ending for the committer. (shrink)

We present RAWLSNET, a system for altering Bayesian Network (BN) models to satisfy the Rawlsian principle of fair equality of opportunity (FEO). RAWLSNET's BN models generate aspirational data distributions: data generated to reflect an ideally fair, FEO-satisfying society. FEO states that everyone with the same talent and willingness to use it should have the same chance of achieving advantageous social positions (e.g., employment), regardless of their background circumstances (e.g., socioeconomic status). Satisfying FEO requires alterations to social structures such (...) as school assignments. Our paper describes RAWLSNET, a method which takes as input a BN representation of an FEO application and alters the BN's parameters so as to satisfy FEO when possible, and minimize deviation from FEO otherwise. We also offer guidance for applying RAWLSNET, including on recognizing proper applications of FEO. We demonstrate the use of our system with publicly available data sets. RAWLSNET's altered BNs offer the novel capability of generating aspirational data for FEO-relevant tasks. Aspirational data are free from the biases of real-world data, and thus are useful for recognizing and detecting sources of unfairness in machine learning algorithms besides biased data. (shrink)

The Recursive Bayesian Net formalism was originally developed for modelling nested causal relationships. In this paper we argue that the formalism can also be applied to modelling the hierarchical structure of mechanisms. The resulting network contains quantitative information about probabilities, as well as qualitative information about mechanistic structure and causal relations. Since information about probabilities, mechanisms and causal relations is vital for prediction, explanation and control respectively, an RBN can be applied to all these tasks. We show in (...) particular how a simple two-level RBN can be used to model a mechanism in cancer science. The higher level of our model contains variables at the clinical level, while the lower level maps the structure of the cell's mechanism for apoptosis. (shrink)

When we talk about the coherence of a story, we seem to think of how well its individual pieces fit together—how to explicate this notion formally, though? We develop a Bayesian network based coherence measure with implementation in _R_, which performs better than its purely probabilistic predecessors. The novelty is that by paying attention to the network structure, we avoid simply taking mean confirmation scores between all possible pairs of subsets of a narration. Moreover, we assign special (...) importance to the weakest links in a narration, to improve on the other measures’ results for logically inconsistent scenarios. We illustrate and investigate the performance of the measures in relation to a few philosophically motivated examples, and (more extensively) using the real-life example of the Sally Clark case. (shrink)

One of the most challenging tasks when adopting Bayesian networks is the one of learning their structure from data. This task is complicated by the huge search space of possible solutions and by the fact that the problem isNP-hard. Hence, a full enumeration of all the possible solutions is not always feasible and approximations are often required. However, to the best of our knowledge, a quantitative analysis of the performance and characteristics of the different heuristics to solve this problem (...) has never been done before. For this reason, in this work, we provide a detailed comparison of many different state-of-the-art methods for structural learning on simulated data considering both BNs with discrete and continuous variables and with different rates of noise in the data. In particular, we investigate the performance of different widespread scores and algorithmic approaches proposed for the inference and the statistical pitfalls within them. (shrink)