A new reverse reduce-error ensemble pruning algorithm

Although greedy algorithms possess high efficiency, they often receive suboptimal solutions of the ensemble pruning problem, since their exploration areas are limited in large extent. And another marked defect of almost all the currently existing ensemble pruning algorithms, including greedy ones, consists in: they simply abandon all of the classifiers which fail in the competition of ensemble selection, causing a considerable waste of useful resources and information. Inspired by these observations, an interesting greedy Reverse Reduce-Error (RRE) pruning algorithm incorporated with the operation of subtraction is proposed in this work. The RRE algorithm makes the best of the defeated candidate networks in a way that, the Worst Single Model (WSM) is chosen, and then, its votes are subtracted from the votes made by those selected components within the pruned ensemble. The reason is because, for most cases, the WSM might make mistakes in its estimation for the test samples. And, different from the classical RE, the near-optimal solution is produced based on the pruned error of all the available sequential subensembles. Besides, the backfitting step of RE algorithm is replaced with the selection step of a WSM in RRE. Moreover, the problem of ties might be solved more naturally with RRE. Finally, soft voting approach is employed in the testing to RRE algorithm. The performances of RE and RRE algorithms, and two baseline methods, i.e., the method which selects the Best Single Model (BSM) in the initial ensemble, and the method which retains all member networks of the initial ensemble (ALL), are evaluated on seven benchmark classification tasks under different initial ensemble setups. The results of the empirical investigation show the superiority of RRE over the other three ensemble pruning algorithms.     

Indicators for inspection and maintenance planning of concrete structures

Based on an idea introduced by Benjamin and Cornell (1970. Probability, statistics and decision for civil engineers. New York: McGaw Hill) and previous works by the authors it is demonstrated how condition indicators may be formulated for the general purpose of quality control and for assessment and inspection planning in particular. The formulation facilitates quality control based on sampling of indirect information about the condition of the considered components. This allows for a Bayesian formulation of the indicators whereby the experience and expertise of the inspection personnel may be fully utilized and consistently updated as frequentistic information is collected. The approach is illustrated on an example considering a concrete structure subject to corrosion. It is shown how half-cell potential measurements may be utilized to update the probability of excessive repair after 50 years. Furthermore in the same example it is shown how the concept of condition indicators might be applied to develop a cost optimal maintenance strategy composed of preventive and corrective repair measures.     

A Bayesian approach to the multivariate Behrens-Fisher problem under the assumption of proportional covariance matrices

Summary Two independent random samples of sizesN ₁ andN ₂ from multivariate normal populationsN _p (θ₁,∑₁) andN _p (θ₂,∑₂) are considered. Under the null hypothesisH ₀: θ₁=θ₂, a single θ is generated from aN _p(μ, Σ) prior distribution, while underH ₁: θ₁≠θ₂ two means are generated from the exchangeable priorN _p(μ,σ). In both cases Σ will be assumed to have a vague prior distribution. For a simple covariance structure, the Bayes factorB and minimum Bayes factor in favour of the null hypotheses is derived. The Bayes risk for each hypothesis is derived and a strategy is discussed for using the Bayes factor and Bayes risks to test the hypothesis.     

Bayes estimators for reliability measures in geometric distribution model using masked system life test data

This paper presents Bayes estimators for the reliability measures of the individual components in a multi-component systems in the presence of masked system life test data. The life time distributions of the system components are assumed to be geometric with different parameters. Two-sided Bayesian probability intervals of the parameters are also derived. Numerical simulation study is given in order to: (i) explain how one can apply the theoretical results obtained, (ii) study the influence of the sample size and masking level on the accuracy of point estimates.     

Information-Based Evaluation Criterion for Classifier's Performance

In the past few years many systems for learning decision rules from examples were developed. As different systems allow different types of answers when classifying new instances, it is difficult to appropriately evaluate the systems' classification power in comparison with other classification systems or in comparison with human experts. Classification accuracy is usually used as a measure of classification performance. This measure is, however, known to have several defects. A fair evaluation criterion should exclude the influence of the class probabilities which may enable a completely uninformed classifier to trivially achieve high classification accuracy. In this paper a method for evaluating the information score of a classifier's answers is proposed. It excludes the influence of prior probabilities, deals with various types of imperfect or probabilistic answers and can be used also for comparing the performance in different domains.     

Non-parametric empirical Bayes procedure

In this paper we introduce an empirical Bayes procedure for estimating an unknown parameter, say θ. This procedure gives the empirical Bayes estimator for θ and its associated minimum posterior risk in closed forms without estimating the unknown prior density function of θ. In such procedure the posterior probability density function of θ is not required. A sufficient statistic for θ with conditional probability density function in the one parameter exponential family is required. Instead of estimating the unknown prior density function, the marginal density function of the sufficient statistic must be estimated. As special cases the empirical Bayes estimators and their respective minimum posterior risks of the failure rate for the exponential distribution, the unknown scale parameters of Weibull and gamma distributions are obtained in simple forms as special cases. Numerical results and a simulation study are introduced to (i) investigate how the number of available past experiments and the sample size of each influence the accuracy of the empirical Bayes estimator, (ii) make a comparison between the presented procedure and the Bayes procedure when the prior probability density function of the parameter θ is gamma.     

Can being scared cause tummy aches? Naive theories, ambiguous evidence, and preschoolers' causal inferences.

Causal learning requires integrating constraints provided by domain-specific theories with domain-general statistical learning. In order to investigate the interaction between these factors, the authors presented preschoolers with stories pitting their existing theories against statistical evidence. Each child heard 2 stories in which 2 candidate causes co-occurred with an effect. Evidence was presented in the form: AB→E; CA→E; AD→E; and so forth. In 1 story, all variables came from the same domain; in the other, the recurring candidate cause, A, came from a different domain (A was a psychological cause of a biological effect). After receiving this statistical evidence, children were asked to identify the cause of the effect on a new trial. Consistent with the predictions of a Bayesian model, all children were more likely to identify A as the cause within domains than across domains. Whereas 3.5-year-olds learned only from the within-domain evidence, 4- and 5-year-olds learned from the cross-domain evidence and were able to transfer their new expectations about psychosomatic causality to a novel task. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

改进贝叶斯分类的智能短信分类方法

随着移动通信技术的不断发展,手机的普及率在不断上升,而短信作为传统的移动通信服务,长久以来一直在人们的日常生活中占据着极为重要的位置。可以说,短信在一定程度上记录了人们生活的轨迹。但是,现有的短信管理系统仅对短信进行以联系人为特征分类、以时间为顺序显示的简单非智能化的管理,导致了用户手机中各类短信混杂不清,短信的管理效率极低。通过研究短信的特征,分析传统的基于文档频率的特征值提取方法和基于互信息的特征值提取方法的优势与不足,提出了一种适用于短信的基于词频和互信息的特征值提取方法,并结合短信长度实现了一种改进的贝叶斯分类算法。实验证明,算法在进行短信分类时可以得到相当可观的召回率和准确率。