基于贝叶斯网络模型的信息检索

贝叶斯网络模型是解决信息检索领域问题的有效手段，因为它可以表示术语间的条件概率和概念语义，并依此预测用户查询和文档间的相似度。本文对信息检索中的贝叶斯网络模型做了若干改进，加入了一些新特征，并用实验结果证明了通过提高贝叶斯网络模型的质量和调整其中的参数，我们的方法取得了较好的检索效果。     

Approximate importance sampling Monte Carlo for data assimilation

Importance sampling Monte Carlo offers powerful approaches to approximating Bayesian updating in sequential problems. Specific classes of such approaches are known as particle filters. These procedures rely on the simulation of samples or ensembles of the unknown quantities and the calculation of associated weights for the ensemble members. As time evolves and/or when applied in high-dimensional settings, such as those of interest in many data assimilation problems, these weights typically display undesirable features. The key difficulty involves a collapse toward approximate distributions concentrating virtually all of their probability on an implausibly few ensemble members.

After reviewing ensembling, Monte Carlo, importance sampling and particle filters, we present some approximations intended to moderate the problem of collapsing weights. The motivations for these suggestions are combinations of (i) the idea that key dynamical behavior in many systems actually takes place on a low dimensional manifold, and (ii) notions of statistical dimension reduction. We illustrate our suggestions in a problem of inference for ocean surface winds and atmospheric pressure. Real observational data are used.     

高速公路车辆行驶路径模糊匹配分析

高速公路车辆行驶路径在省界站处被自然分隔为两段,不能真实反映行驶车辆在高速公路上的路径信息.本文选取收费系统中车牌号、出入口时间、轴型、车货总重等字段信息,利用贝叶斯方法对分隔路径进行模糊匹配,从而得出行驶车辆在高速公路的全路径信息.     

NON-PARAMETRIC TOPIC MODEL FOR DISCOVERING GEOGRAPHICAL TOPIC VARIATIONS

This paper presents a non-parametric topic model that captures not only the latent topics in text collections, but also how the topics change over space. Unlike other recent work that relies on either Gaussian assumptions or diseretization of locations, here topics are associated with a distance dependent Chinese Restaurant Process （ddCRP）, and for each document, the observed words are influenced by the document＇s GPS-tag. Our model allows both unbound number and flexible distribution of the geographical variations of the topics＇ content. We develop a Gibbs sampler for the proposal, and compare it with existing models on a real data set basis.     

Bayesian sequential reliability for Weibull and related distributions

Assume that the probability density function for the lifetime of a newly designed product has the form: [H(t)/Q()] exp{–H(t)/Q()}. The Exponential(), Rayleigh, WeibullW(, ) and Pareto pdf's are special cases.Q() will be assumed to have an inverse Gamma prior. Assume thatm independent products are to be tested with replacement. A Bayesian Sequential Reliability Demonstration Testing plan is used to eigher accept the product and start formal production, or reject the product for reengineering. The test criterion is the intersection of two goals, a minimal goal to begin production and a mature product goal. The exact values of various risks and the distribution of total number of failures are evaluated. Based on a result about a Poisson process, the expected stopping time for the exponential failure time is also found. Included in these risks and expected stopping times are frequentist versions, thereof, so that the results also provide frequentist answers for a class of interesting stopping rules.This research was supported by NSF grants DMS-8703620 and DMS-8923071, and forms part of the Ph.D. Thesis of the first author, the development of which was supported in part by a David Ross grant at Purdue University. The authors thank the editors and a referee for insightful comments and suggestions.     

Assessing the predictive effectiveness of the variety seeking and reinforcement models

Prediction of customer choice behaviour has been a big challenge for marketing researchers. They have adopted various models to represent customers purchase patterns. Some researchers considered simple zero–order models. Others proposed higher–order models to represent explicitly customers tendency to seek [variety] or [reinforcement] as they make repetitive choices. Nevertheless, the question [Which model has the highest probability of representing some future data?] still prevails. The objective of this paper is to address this question. We assess the predictive effectiveness of the well–known customer choice models. In particular, we compare the predictive ability of the [dynamic attribute satiation] (DAS) model due to McAlister (Journal of Consumer Research, 91, pp. 141–150, 1982) with that of the well–known stochastic variety seeking and reinforcement behaviour models. We found that the stochastic [beta binomial] model has the best predictive effectiveness on both simulated and real purchase data. Using simulations, we also assessed the effectiveness of the stochastic models in representing various complex choice processes generated by the DAS. The beta binomial model mimicked the DAS processes the best. In this research we also propose, for the first time, a stochastic choice rule for the DAS model.     

Bayesian compressive sensing for cluster structured sparse signals

In traditional framework of compressive sensing (CS), only sparse prior on the property of signals in time or frequency domain is adopted to guarantee the exact inverse recovery. Other than sparse prior, structures on the sparse pattern of the signal have also been used as an additional prior, called model-based compressive sensing, such as clustered structure and tree structure on wavelet coefficients. In this paper, the cluster structured sparse signals are investigated. Under the framework of Bayesian compressive sensing, a hierarchical Bayesian model is employed to model both the sparse prior and cluster prior, then Markov Chain Monte Carlo (MCMC) sampling is implemented for the inference. Unlike the state-of-the-art algorithms which are also taking into account the cluster prior, the proposed algorithm solves the inverse problem automatically—prior information on the number of clusters and the size of each cluster is unknown. The experimental results show that the proposed algorithm outperforms many state-of-the-art algorithms.     

电信反欺诈系统(AFS)的设计与实现

AFS系统(Anti-FraudSystem)通过运用数据挖掘技术中的贝叶斯分类器(BayesianClassifiers)来解决电信行业中的客户欺诈问题。首先介绍了开发电信反欺诈系统的必要性和紧迫性,论述了采用贝叶斯分类器的理由,然后介绍了AFS系统的构架和功能,并着重阐述了系统中采用贝叶斯分类器实现反欺诈功能的具体思路,最后对AFS系统进行了总体评价。     

基于贝叶斯算法的XG-EPON业务感知机制

提出了一种基于贝叶斯算法的XG-PON业务感知机制,该机制采用贝叶斯算法对业务进行分类,以实现高效的业务感知。在此基础上通过OLT与ONU的交互,优化XG-PON的资源分配。对基于贝叶斯算法的XG-PON业务感知机制进行仿真,仿真结果表明该机制能有效降低丢包率并缩短时延,提高了系统对多业务的支持能力。