We consider the problem of model selection and accounting for model uncertainty in high-dimensional contingency tables, motivated by expert system applications. The approach most used currently is a stepwise strategy guided by tests based on approximate asymptotic P-values leading to the selection of a single model; inference is then conditional on the selected model. The sampling properties of such a strategy are complex, and the failure to take account of model uncertainty leads to underestimation of uncertainty about quantities of interest. In principle, a panacea is provided by the standard Bayesian formalism which averages the posterior distributions of the quantity of interest under each of the models, weighted by their posterior model probabilities. Furthermore, this approach is optimal in the sense of maximising predictive ability. However, this has not been used in practice because computing the posterior model probabilities is hard and the number of models is very large (often greater than 1011). We argue that the standard Bayesian formalism is unsatisfactory and we propose an alternative Bayesian approach that, we contend, takes full account of the true model uncertainty byaveraging overamuch smaller set of models. An efficient search algorithm is developed for nding these models. We consider two classes of graphical models that arise in expert systems: the recursive causal models and the decomposable

We describe the development of a monitoring system which uses sensor observation data about discrete events to construct dynamically a probabilistic model of the world. This model is a Bayesian network incorporating temporal aspects, which we call a Dynamic Belief Network; it is used to reason under uncertainty about both the causes and consequences of the events being monitored. The basic dynamic construction of the network is data-driven. However the model construction process combines sensor data about events with externally provided information about agents' behaviour, and knowledge already contained within the model, to control the size and complexity of the network. This means that both the network structure within a time interval, and the amount of history and detail maintained, can vary over time. We illustrate the system with the example domain of monitoring robot vehicles and people in a restricted dynamic environment using light-beam sensor data. In addition to presenting a ...

In this paper we present a ne_(l n [!ii ' with respect to Bayesian networks con- ogy for inducing Bayesian network structures frop3 titute the roblem of the evidence propagation and a database of cases. The methodology is based oap&lll searching for the best ordering of the system vari- the problem of the model search. The problem of shies by means of genetic algorithl{. Since his th_vidence propagation consists of once the vMproblem of finding an optimal ordea. teeuarue}rables are known, the assignment of resembles the traveling salesman p'FolUleh)ve use .... IW. ....... probablhles to the values of the rest of the van genetic operators that were developed for the latter - problem. The quality of a variable ordering is eval- ables. Cooper [4] demonstrated that this problem Mated with the algorithm K2. We present empirical results that were obtained with a simulation of the ALARM network.

Acyclic digraphs (ADGs) are widely used to describe dependences among variables in multivariate distributions. In particular, the likelihood functions of ADG models admit convenient recursive factorizations that often allow explicit maximum likelihood estimates and that are well suited to building Bayesian networks for expert systems. There may, however, be many ADGs that determine the same dependence (= Markov) model. Thus, the family of all ADGs with a given set of vertices is naturally partitioned into Markov-equivalence classes, each class being associated with a unique statistical model. Statistical procedures, such as model selection or model averaging, that fail to take into account these equivalence classes, may incur substantial computational or other inefficiencies. Recent results have shown that each Markov-equivalence class is uniquely determined by a single chain graph, the essential graph, that is itself Markov-equivalent simultaneously to all ADGs in the equivalence clas...

A framework for log-linear models with context specific independence structures, i.e. conditional independencies holding only for specific values of the conditioning variables is introduced. This framework is constituted by the class of split models. Also a software package named YGGDRASIL which is designed for statistical inference in split models is presented. Split models are an extension of graphical models for contingency tables. The treatment of split models includes estimation, representation and a Markov property for reading off independencies holding in a specific context. Two examples, including an illustration of the use of YGGDRASIL are

The theoretical background for a program for establishing expert systems on the basis of observations and expert knowledge is presented. Block recursive models form the basis of the statistical modelling. These models, together with various model selection methods for automatic model selection, are presented. Additionally, the connection between a block recursive model and expert systems based on causal probabilistic networks is treated. A medical example concerning diagnosis of coronary artery disease forms the basis for an evaluation of the expert systems established. Keywords: causal probabilistic networks, graphical association models, machine learning, model selection, selection criteria, selection strategies. 1 Introduction BIFROST is a program for semi-automatic knowledge acquisition and is a continuation developments made in (Greve, Hjsgaard, Skjth and Thiesson 1990). The objective is to obtain preliminary causal models for use in the HUGIN expert system shell (Ander...

This paper is concerned with handling uncertainty as part of the analysis of data from a medical study. The study is investigating connections between the birth weight of babies and the dietary intake of their mothers. Bayesian belief networks were used in the analysis. Their perceived benefits include (i) an ability to represent the evidence emerging from the evolving study, dealing effectively with the inherent uncertainty involved; (ii) providing a way of representing evidence graphically to facilitate analysis and communication with clinicians; (iii) helping in the exploration of the data to reveal undiscovered knowledge; and (iv) providing a means of developing an expert system application.