Deduction from Conditional Knowledge on Bayesian Networks with Interval Probability Parameters

Interval data are widely used in real applications to represent the values of quantities in uncertain situations. However, the implied probabilistic causal relationships among interval-valued variables with interval data cannot be represented and inferred by general Bayesian networks with point-based probability parameters. Thus, it is desired to extend the general Bayesian network with effective mechanisms of representation, learning and inference of probabilistic causal relationships implied in interval data. In this paper, we define the interval probabilities, the bound-limited weak conditional interval probabilities and the probabilistic description, as well as the multiplication rules. Furthermore, we propose the method for learning the Bayesian network structure from interval data and the algorithm for corresponding approximate inferences. Experimental results show that our methods are feasible, and we conclude that the Bayesian network with interval probability parameters is the expansion of the general Bayesian network.

Conditional event algebra (CEA) was developed in order to represent conditional probabilities with differing antecedents by the probability evaluation of well-defined individual “conditional” events in a single larger space extending the original unconditional one. These conditional events can then be combined logically before being evaluated. A major application of CEA is to data fusion problems, especially the testing of hypotheses concerning the similarity or redundancy among inference rules through use of probabilistic distance functions which critically require probabilistic conjunctions of conditional events. Relational event algebra (REA) is a further extension of CEA, whereby functions of probabilities formally representing single event probabilities — not just divisions as in the case of CEA — are shown to represent actual “relational” events relative to appropriately determined larger probability spaces. Analogously, utilizing the logical combinations of such relational events allows for testing of hypotheses of similarity between data fusion models represented by functions of probabilities. Independent of, and prior to this work, it was proven that a major portion of fuzzy logic — a basic tool for treating natural language descriptions — can be directly related to probability theory via the use of one point random set coverage functions. In this paper, it is demonstrated that a natural extension of the one point coverage link between fuzzy logic and random set theory can be used in conjunction with CEA and REA to test for similarity of natural language descriptions.

Bayesian networks (BNs) are being studied in recent years for system diagnosis, reliability analysis, and design of complex engineered systems. In several practical applications, BNs need to be learned from available data before being used for design or other purposes. Current BN learning algorithms are mainly developed for networks with only discrete variables. Engineering design problems often consist of both discrete and continuous variables. This paper develops a framework to handle continuous variables in BN learning by integrating learning algorithms of discrete BNs with Gaussian mixture models (GMMs). We first make the topology learning more robust by optimizing the number of Gaussian components in the univariate GMMs currently available in the literature. Based on the BN topology learning, a new multivariate Gaussian mixture (MGM) strategy is developed to improve the accuracy of conditional probability learning in the BN. A method is proposed to address this difficulty of MGM modeling with data of mixed discrete and continuous variables by mapping the data for discrete variables into data for a standard normal variable. The proposed framework is capable of learning BNs without discretizing the continuous variables or making assumptions about their conditional probability densities (CPDs). The applications of the learned BN to uncertainty quantification and model calibration are also investigated. The results of a mathematical example and an engineering application example demonstrate the effectiveness of the proposed framework.

Bayesian networks are often proposed as a method for high-level information fusion. However, a Bayesian network relies on strong assumptions about the underlying probabilities. In many cases it is not realistic to require such precise probability assessments. We show that there exists a significant set of problems where credal networks outperform Bayesian networks, thus enabling more dependable decision making for this type of problems. A credal network is a graphical probabilistic method that utilizes sets of probability distributions, e.g., interval probabilities, for representation of belief. Such a representation allows one to properly express epistemic uncertainty, i.e., uncertainty that can be reduced if more information becomes available. Since reducing uncertainty has been proposed as one of the main goals of information fusion, the ability to represent epistemic uncertainty becomes an important aspect in all fusion applications.

In this thesis, we concentrate on the relationship of probability intervals and random sets. Furthermore, we are interested in solving uncertain linear programming problems with probability interval and random set parameters. We discover the conditions to verify when a given probability interval obtains the same information as a random set information. If these conditions are satisfied, we can transform a problem that contains both types of uncertainty into a problem which has only the random set information. In addition, we use an idea from decision making theory with random sets for solving this problem. If a probability interval does not satisfy these conditions, we can solve the problem for finding the optimistic and pessimistic expected recourse values. Finally, we present an algorithm for checking these conditions and constructing appropriate distributions for each of the optimistic and pessimistic approaches.

Bayesian networks (BNs) model problems that involve uncertainty. A BN is a directed graph, whose nodes are the uncertain variables and whose edges are the causal or influential links between the variables. Associated with each node is a set of conditional probability functions that model the uncertain relationship between the node and its parents. The benefits of using BNs to model uncertain domains are well known, especially since the recent breakthroughs in algorithms and tools to implement them. However, there have been serious problems for practitioners trying to use BNs to solve realistic problems. This is because, although the tools make it possible to execute large-scale BNs efficiently, there have been no guidelines on building BNs. Specifically, practitioners face two significant barriers. The first barrier is that of specifying the graph structure such that it is a sensible model of the types of reasoning being applied. The second barrier is that of eliciting the conditional probability values. In this paper we concentrate on this first problem. Our solution is based on the notion of generally applicable “building blocks”, called idioms, which serve solution patterns. These can then in turn be combined into larger BNs, using simple combination rules and by exploiting recent ideas on modular and object oriented BNs (OOBNs). This approach, which has been implemented in a BN tool, can be applied in many problem domains. We use examples to illustrate how it has been applied to build large-scale BNs for predicting software safety. In the paper we review related research from the knowledge and software engineering literature. This provides some context to the work and supports our argument that BN knowledge engineers require the same types of processes, methods and strategies enjoyed by systems and software engineers if they are to succeed in producing timely, quality and cost-effective BN decision support solutions.

This paper develops a model validation method in the case of the full scale tests for the system model are infeasible. The Bayesian network with uncertain conditional probability parameters is used to represent the relations between the large computational model and its smaller modules. The interval probability theory is adopted to extrapolate the posterior probability of the interested variable in the uncertain Bayesian network. An interval valued Bayes factor is obtained to be the metric for model validation.

The potential applications and challenges of uncertain data mining have recently attracted interests from researchers. Most uncertain data mining algorithms consider aleatory (random) uncertainty of data, i.e. these algorithms require that exact probability distributions or confidence values are attached to uncertain data. However, knowledge about uncertainty may be incomplete in the case of epistemic (incomplete) uncertainty of data, i.e. probabilities of uncertain data may be imprecise, coarse, or missing in some applications. The paper focuses on uncertain data which miss probabilities, specially, value-uncertain discrete objects which miss probabilities (for short uncertain objects). On the other hand, classification is one of the most important tasks in data mining. But, to the best of our knowledge, there is no method to learn Naïve Bayesian classifier from uncertain objects. So the paper studies Naïve Bayesian classification of uncertain objects. Firstly, the paper defines interval probabilities of uncertain objects from probabilistic cardinality point of view, and bridges the gap between uncertain objects and the theory of interval probability by proving that interval probabilities are F-probabilities. Secondly, based on the theory of interval probability, the paper defines conditional interval probabilities including the intuitive concept and the canonical concept, and the conditional independence of the intuitive concept. Further, the paper gives a formula to effectively compute the intuitive concept. Thirdly, the paper presents a Naïve Bayesian classifier with interval probability parameters which can handle both uncertain objects and certain objects. Finally, experiments with uncertain objects based on UCI data show satisfactory performances.

During the last decades, "Transport Demand" and "Mobility" has been a continuously developing branch in the transport literature.This is reflected in the great amount of research papers published in scientific magazines dealing with trip matrix estimation (see (Doblas & Benítez, 2005)) and traffic assignment problems (see (Praskher & Bekhor, 2004)).Current traffic models reproduce the mobility using several data inputs, in particular prior trip matrix, link counts, etc. (see (Yang & Zhou, 1998)) which are data only from a subset of the problem variables, and its size will depend on the available budget for the study being carried on.Among the problems faced for solving these models we can emphasize the high number of possible solutions which is usually solved by choosing the solution where the model results best fits real data.Nevertheless a model does not have to reproduce only the real data, but also must reproduce accurately all the variables.To this end, the aim of this paper consists of presenting two Bayesian network models for traffic estimation, trying to bring a new tool to the transportation field.The first one deals with the problem of link flows, trip matrix estimation and traffic counting location and in the second one we propose a Bayesian network for route flow estimation (and hence link flows and OD flows) using data from plate scanning technique together with a model for optimal plate scanning device location.Since a Gaussian Bayesian network is used, these models allow us to update the predictions from a small subset of real data and probability intervals or regions are obtained to get an idea of the associated uncertainties.In addition dealing with data from the plate scanning approach we improve the under-specification level of the traffic flow estimation problem. Some background on Bayesian network modelsA Bayesian network (see, (Castillo et al., 1999)) is a pair (G, P ), where G is a directed acyclic graph (DAG) defined on a set of nodes X (the random variables), and P = {p(x 1 |π 1 ), . . ., p(x n |π n )} is a set of n conditional probability densities (CPD), one for each variable, and Π i is the set of parents of node X i in G.The set P defines the associated joint 19 www.intechopen.comBayesian Network 318 probability density of all nodes as p(1)The graph G contains all the qualitative information about the relationships among the variables, no matter which probability values are assigned to them 1 .Complementary, the probabilities in P contain the quantitative information, i.e., they complement the qualitative properties revealed by the graphical structure.

The assessments for the various conditional probabilities of a Bayesian belief network inevitably are inaccurate, influencing the reliability of its output. By subjecting the network to a i>sensitivity analysis with respect to its conditional probabilities, the reliability of its output can be investigated. Unfortunately, straightforward sensitivity analysis of a belief network is highly time-consuming. In this paper, we show that by qualitative considerations several analyses can be identified as being uninformative as the conditional probabilities under study cannot affect the output. In addition, we show that the analyses that are informative comply with simple mathematical functions. More specifically, we show that a belief network's output can be expressed as a quotient of two functions that are linear in a conditional probability under study. These properties allow for considerably reducing the computational burden of sensitivity analysis of Bayesian belief networks.

EXP’s IC3D system classifies SAR or LiDAR derived 3D point clouds using a deep representation learning approach, producing as output a vector of categorical posterior probabilities of target classifications. Such posterior probabilities are suitable observational inputs to a Bayesian belief network (BBN), such as the EXP Shadow Compass system. In concert with conditional probabilities of intermediate events depending on the observation states, the Bayesian network computes posterior probabilities for events conditionally dependent on those intermediate event states. We demonstrate this approach by computing posterior event probabilities for sample analyst scenarios with intermediate event conditional probabilities specified by analysts. Future work could include extending the Bayesian network approach to discovery of the network topology from analyst data.

Bayesian networks (BN) are an excellent tool for modeling uncertainties in systems with several interdependent variables. A BN is a directed acyclic graph, and consists of a structure, or the set of directional links between variables that depend on other variables, and conditional probabilities (CP) for each variable. In this project, we apply BN's to understand uncertainties in NIF ignition experiments. One can represent various physical properties of National Ignition Facility (NIF) capsule implosions as variables in a BN. A dataset containing simulations of NIF capsule implosions was provided. The dataset was generated from a radiation hydrodynamics code, and it contained 120 simulations of 16 variables. Relevant knowledge about the physics of NIF capsule implosions and greedy search algorithms were used to search for hypothetical structures for a BN. Our preliminary results found 6 links between variables in the dataset. However, we thought there should have been more links between the dataset variables based on the physics of NIF capsule implosions. Important reasons for the paucity of links are the relatively small size of the dataset, and the sampling of the values for dataset variables. Another factor that might have caused the paucity of links is the fact that in the dataset, 20% of the simulations represented successful fusion, and 80% didn't, (simulations of unsuccessful fusion are useful for measuring certain diagnostics) which skewed the distributions of several variables, and possibly reduced the number of links. Nevertheless, by illustrating the interdependencies and conditional probabilities of several parameters and diagnostics, an accurate and complete BN built from an appropriate simulation set would provide uncertainty quantification for NIF capsule implosions.

Uncertainty analyses in fault trees and Bayesian networks using FORM/SORM methods

INTRODUCTION: As endovascular treatment options for acute ischemic stroke evolve, it is important to study the conditions and decisions that lead to successful outcomes so that future interventions may be better tailored for each individual. Bayesian networks provide a method for studying the conditional dependencies between variables and have been applied in other areas of medicine. In this work, we present a Bayesian network of acute ischemic stroke for analyzing the dependencies between a patient’s clinical and imaging presentation, therapeutic interventions, and outcomes. METHODS: 790 unique episodes of acute ischemic stroke from the last five years (2006-2012) were retrieved from our institution’s quality improvement repository. A subset of variables from each case was extracted and modeled in a Bayesian network. Variable selection and connectivity was guided by a review of current practice guidelines and the domain knowledge of clinical investigators. Conditional probabilities between variables were then calculated using expectation maximization. RESULTS: The Bayesian network may be manipulated through a graphical user interface to investigate the likelihood of different clinical scenarios. For example, evidence regarding patient presentation may be set and then combinations of interventions may be applied to observe possible outcomes. Additionally, the model accommodates the integration of utility nodes to support exploration of a patient’s projected quality-adjusted life expectancy (QALE) as a function of their clinical state and variable treatment decisions. CONCLUSION: The presented model supports a pertinent set of variables and provides an initial tool for clinicians and researchers to study how combinations of patient evidence and interventions affect outcomes in acute stroke. This information may be useful for suggesting subsequent investigations on methods for improving existing treatment protocols.

Use of evidential reasoning for eliciting bayesian subjective probabilities in human reliability analysis: A maritime case