Lymphoepithelial Cyst in the Posterior Tongue: a Case Report (original) (raw)
Related papers
Generalizing e-Bay.NET: An Approach to Recommendation Based on Probabilistic Computing
2005
In this paper, we shall present the theoretical developments related to extending existing e-Bay.NET recommendation system in order to improve its expressiveness. In particular, we shall make them more flexible and more general by enabling it to handle evidence items with a finer granularity so that more accurate information may be obtained when user preferences are elicited. The model is based on the formalism of Bayesian networks, and this extension requires the design of new methods to estimate conditional probability distributions and also a new algorithm to compute the posterior probabilities of relevance.
A Probabilistic Graphical Model for Mallows Preferences
2016
Reasoning about preference distributions is an important task in various areas (e.g., recommender systems, social choice), and is critical for learning the parameters of the distribution. We consider the Mallows model with arbitrary pairwise comparisons as evidence. Existing inference methods are able to reason only with evidence that abides to a restrictive form. We establish the conditional independences in the Mallows model, and apply them to develop a Bayesian network that enables querying and sampling from the Mallows posterior (the conditional probability space with the evidence incorporated). While inference over the Mallows posterior is computationally hard in general, our translation allows to utilize the wealth of tools for inference over Bayesian networks. Moreover, we show how our translation gives rise to new results on significant cases with a polynomial-time inference.
CRI eBooks, 2017
Accurate prediction of future claims is a fundamentally important problem in insurance. The Bayesian approach is natural in this context, as it provides a complete predictive distribution for future claims. The classical credibility theory provides a simple approximation to the mean of that predictive distribution as a point predictor, but this approach ignores other features of the predictive distribution, such as spread, that would be useful for decision making. In this article, we propose a Dirichlet process mixture of log-normals model and discuss the theoretical properties and computation of the corresponding predictive distribution. Numerical examples demonstrate the benefit of our model compared to some existing insurance loss models, and an R code implementation of the proposed method is also provided.
2005
Although Bayesian Nets (BNs) are increasingly being used to solve real world risk problems, their use is still constrained by the difficulty of constructing the node probability tables (NPTs) for each node. In the absence of hard data, we must rely on domain experts to provide, often subjective, judgements to inform the NPTs. A key challenge is to construct relevant NPTs using the minimal amount of expert elicitation, recognising that it is rarely cost-effective to elicit complete sets of probability values. We describe a simple approach to defining NPTs for a large class of commonly occurring nodes (called ranked nodes). The approach is based on the doubly truncated Normal distribution with a central tendency that is invariably a type of weighted function of the parent nodes. In extensive real-world case studies we have found that this approach is sufficient for generating the NPTs of a very large class of nodes. The approach has been automated and is thus accessible to all types of domain experts, including those with little statistical expertise. The result has been that such individuals have been able to build large-scale realistic BN models that solve important problems. Hence, this work represents a breakthrough in BN research and technology since it can make the difference between being able to build realistic BN models and not.
Bayesian Statistics and Marketing
Marketing Science, 2003
Bayesian methods have become widespread in the marketing literature. We review the essence of the Bayesian approach and explain why it is particularly useful for marketing problems. While the appeal of the Bayesian approach has long been noted by researchers, recent developments in computational methods and expanded availability of detailed marketplace data has fueled the Bayesian growth in marketing. We emphasize the modularity and flexibility of modern Bayesian approaches. Finally, the usefulness of Bayesian methods in situations in which there is limited information about a large number of units or where the information comes from different sources is noted
Modelling preference data with the Wallenius distribution
Journal of the Royal Statistical Society: Series A (Statistics in Society)
The Wallenius distribution is a generalisation of the Hypergeometric distribution where weights are assigned to balls of different colours. This naturally defines a model for ranking categories which can be used for classification purposes. Since, in general, the resulting likelihood is not analytically available, we adopt an approximate Bayesian computational (ABC) approach for estimating the importance of the categories. We illustrate the performance of the estimation procedure on simulated datasets. Finally, we use the new model for analysing two datasets concerning movies ratings and Italian academic statisticians' journal preferences. The latter is a novel dataset collected by the authors.
BugsXLA : Bayes for the Common Man
Journal of Statistical Software, 2005
The absence of user-friendly software has long been a major obstacle to the routine application of Bayesian methods in business and industry. It will only be through widespread application of the Bayesian approach to real problems that issues, such as the use of prior distributions, can be practically resolved in the same way that the choice of significance levels has been in the classical approach; although most Bayesians would hope for a much more satisfactory resolution. It is only relatively recently that any general purpose Bayesian software has been available; by far the most widely used such package is WinBUGS. Although this software has been designed to enable an extremely wide variety of models to be coded relatively easily, it is unlikely that many will bother to learn the language and its nuances unless they are already highly motivated to try Bayesian methods. This paper describes a graphical user interface, programmed by the author, which facilitates the specification of a wide class of generalised linear mixed models for analysis using WinBUGS. The program, BugsXLA (v2.1), is an Excel Add-In that not only allows the user to specify a model as one would in a package such as SAS or S-PLUS, but also aids the specification of priors and control of the MCMC run itself. Inevitably, developing a program such as this forces one to think again about such issues as choice of default priors, parameterisation and assessing convergence. I have tried to adopt currently perceived good practices, but mainly share my approach so that others can apply it and, through constructive criticism, play a small part in the ultimate development of the first Bayesian software package truly useable by the average data analyst.
A Bayesian Approach to Modeling Purchase Frequency
2003
Direct marketers are often faced with the task of ranking, or scoring individual customers in terms of their ex- pected value to the firm. A critical element of their scoring systems is expected frequency of customer interaction. In this paper the authors develop a hierarchical Bayes model of purchase frequency that combines a Poisson like- lihood with a gamma mixing
Using Ranked Nodes to Model Qualitative Judgments in Bayesian Networks
IEEE Transactions on Knowledge and Data Engineering, 2000
Although Bayesian Nets (BNs) are increasingly being used to solve real world risk problems, their use is still constrained by the difficulty of constructing the node probability tables (NPTs). A key challenge is to construct relevant NPTs using the minimal amount of expert elicitation, recognising that it is rarely cost-effective to elicit complete sets of probability values. We describe a simple approach to defining NPTs for a large class of commonly occurring nodes (called ranked nodes). The approach is based on the doubly truncated Normal distribution with a central tendency that is invariably a type of weighted function of the parent nodes. In extensive real-world case studies we have found that this approach is sufficient for generating the NPTs of a very large class of nodes. We describe one such case study for validation purposes. The approach has been fully automated in a commercial tool, called AgenaRisk, and is thus accessible to all types of domain experts. We believe this work represents a useful contribution to BN research and technology since its application makes the difference between being able to build realistic BN models and not.