DIAGNOSTICS FOR EMPIRICAL BAYES MODELS

1.IntroductionInthelilleaxmodels,manykindsofBayesiandiagnosticshavebeenproposedtoassesstheinfluenceofeaChindividualobservationandthelocalinfluenceofminorperturbationsontheposteriordistributioll(see,[l--4])oronthepredictionoffutuxeobservation(see[5]).However,allthosediagnosticscannotbedirectlyappliedtotheempiricalBayes(EB)models(see[6]and[7]).Inthisarticle,westudythediagnosticproblemsinEBmodelssystematically.FirstlyjadeletiondiagllosticisobtainedtoexaminethesensitivityofeachcaseonEBestimate…     

Paretian型超出损失再保险纯保费的贝叶斯极值估计

导出全Paretian模型的参数的贝叶斯估计式，并将参数估计式应用于纯保费的估计，得到了超出损失再保险保单的纯保费的贝叶斯估计公式。作为一个应用，将所得到的估计结果应用于火灾保险和汽车保险数据。最后通过MC模拟说明所构造的模型的稳健性。     

GMDH-Based Outlier Detection Model in Classi?cation Problems

In many practical classification problems, datasets would have a portion of outliers, which could greatly affect the performance of the constructed models. In order to address this issue, we apply the group method of data handing(GMDH) neural network in outlier detection. This study builds a GMDH-based outlier detection(GOD) model. This model first implements feature selection in the training set L using GMDH neural network. Then a new training set L can be obtained by mapping the selected key feature subset. Next, a linear regression model can be constructed in the set L by ordinary least squares estimation. Further, it eliminates a sample from the set L randomly every time, and then rebuilds a linear regression model. Finally, outlier detection is realized by calculating Cook's distance for each sample. Four different customer classification datasets are used to conduct experiments. Results show that GOD model can effectively eliminate outliers, and compared with the five existing outlier detection models, it generally performs significantly better. This indicates that eliminating outliers can effectively enhance classification accuracy of the trained classification model.     

The Existence of Supporting Prices and Economic Growth under Uncertainty

１ＩｎｔｒｏｄｕｃｔｉｏｎＥｃｏｎｏｍｉｃｐｌａｎｎｉｎｇｉｓｔｈｅｐｒｏｂｌｅｍｏｆｓｏｃｉａｌｄｅｃｉｓｉｏｎｏｒｏｐｔｉｍａｌｄｅｔｅｒｍｉｎａｔｉｏｎ．Ｉｔｃａｎｂｅａｄｊｕｓｔｅｄｔｏｓｕｉｔｔｈｅｓｐｅｃｉｆｉｃｎｅｅｄｓａｎｄｐｒｏｂｌ...     

安全第一准则下的动态资产组合选择

考虑连续时间金融市场的最优资产组合选择问题.在Black-Scholes金融市场设置下,利用Roy提出的安全第一(SafetyFirst)准则,导出了最优常数再调整资产组合投资策略的显式表达式.还将文中的结果与Markowitz均值-方差模型的最优资产组合投资策略进行了比较,并给出概念与结论的一些经济学解释和应用.     

金融时间序列分形维估计的小波方法

讨论了金融时间序列的性质,通过实际数据说明,金融时间序列具有两个重要特性——统计自相似性和非平稳性.利用正交小波变换的方法,给出了其分形维的估计方法.最后,实证分析了国内金融市场,并应用此方法分别得出了上证综合指数序列过程和深证成分指数序列过程的分形维.     

基于成败型数据及分组数据指数寿命型元件失效率经典精确最优置信限

设Ａ是一个指数寿命型元件，设其寿命服从参数为λ的指数分布，λ＞０，λ未知．本文研究基于该元件的一次性检测数据及分组数据，用样本点排序法计算元件失效率λ的经典精确最优置信上限，并证明了所得置信上限的最优性．     

INFERENCE ON THE RANK OF THE GROWTH CURVE MODEL USING MODEL SELECTION METHOD

1 IntroductionConsider the growth curve model introduced by Potthoff and Roy[1] and given byy = Aex e, B ~ (o,In @ z) (1.1)where Y: n x p is an obserwtion ~riX, A: n x k is a destw ~rad across individuals, X: q xpis a design ~rad within individuals, e is an tmknoWn parameter mains, and eaCh row of 6 isindependent and idelltically distribllted with mean zero and positive deface covariance mainsZ. Assume that matrices A and X are Of ranks k and q(S P), respectively. In the aPPlicationt…     

具有多重休假和Min(N,V)-策略控制的Geo/G/1离散时间排队

考虑服务员具有多重休假和系统采用min(N,V)-策略控制的离散时间Geo/G/1排队系统,使用全概率分解技术和更新过程理论,研究了系统在任意时刻n+的瞬态队长分布和稳态队长分布,得到了瞬态队长分布的z-变换表达式和稳态队长分布的递推表达式.进一步,得到了系统在时刻点n,n~-和外部观察时刻点的稳态队长分布.特别地,本文直接获得了一些特殊离散时间排队系统相应的结果.最后,通过数值实例阐述了获得便于计算的稳态队长分布的表达式在系统容量设计中的重要价值.     

基于参数学习的GARCH动态无穷活动率Lévy过程的欧式期权定价

在股票价格中引入漂移率、波动率和随机跳跃三种状态， 建立动态状态空间模型， 并通过局部风险中性定价关系（RNVR）推导无套利定价模型. 以非高斯条件ARMA-NGARCH为基准模型， 构建S&P500指数的离散动态Lévy过程， 并基于序贯贝叶斯的参数学习方法， 进行模型估计和期权定价研究. 结果表明： 动态Lévy过程能够联合刻画时变漂移率、条件波动率和无穷活动率等特征， 且贝叶斯方法的引入提高了期权隐含波动率的定价精度. 同时， 无穷活动率模型在期权定价方面具有显著优势. 在五类滤波中， 无损粒子滤波估计精度最高， 速降调和稳态过程（RDTS）的期权定价误差最小， 而非高斯模型在收益率预测方面没有表现出显著的差异.     

基于宏观仿真模拟的公交线网评价模型

对城市公交线网进行分析评价，是城市公交系统规划和公交线网优化的重要组成部分。实际工作中，由于条件限制，很多评价所需数据无法准确获取。本文应用宏观仿真模拟软件TransCAD软件，对公交线网的设施水平和公交线路客运能力水平评价方法进行了探讨，给出了评价指标计算模型。最后，通过对厦门市的城市公交线网规划方案进行实证评价，得出了较好的结果。     

不同借贷利率下的投资组合选择——-基于均值和VaR的效用最大化模型

用VaR代替方差来度量风险,从而把基于均值和方差的效用函数拓展为基于均值和VaR的一般二元效用函数（关于均值递增,关于VaR递减）,进而研究含无风险资产且具有不同借贷利率时投资组合选择的效用最大化模型.利用均值-VaR模型有效边界的性质,得到了一般效用函数下最大效用存在的条件及最优解的本质特征,并给出了求解的具体方法和数值算法.最后作为结论的直接应用和说明,利用中国股票市场数据给出了一个实例分析.     

R&D企业最优技术投资与分红策略研究

本文研究了存在新技术项目时RD企业的最优技术投资与分红策略.假定企业流动资金变化由对偶风险模型描述,企业破产时存在清算价值,目标是最大化破产前的累计红利现值与清算价值现值之和.运用混合奇异控制-最优停时方法,当企业的每笔收益服从指数分布时本文得到RD企业的最优投资与分红策略及最优值函数的解析表达式.最后,我们分析了模型参数对企业最优投资与分红策略的影响,并用效用无偏差理论分析了新技术项目的价值.     

多智能体系统的分布式预测控制器设计

针对多智能体系统的跟踪及编队控制问题,研究其分布式预测控制器的设计方法。考虑时变通讯拓扑情况,在优化问题中附加一项辅助约束,保证了优化问题的递归可行性及多智能体系统的闭环指数稳定性。在每个采样时刻,各智能体根据通讯拓扑实时更新其代价函数及辅助约束,通过求解带有状态、输入及辅助约束的优化问题来计算其控制输入。仿真结果验证了结论的有效性。     

A partial robust optimization approach to inventory management for the offline-to-online problem under different selling prices

This study examines an optimal inventory strategy when a retailer markets a product at different selling prices through a dual-channel supply chain, comprising an online channel and an offline channel. Using the operating pattern of the offline-to-online (O2O) business model, we develop a partial robust optimization (PRO) model. Then, we provide a closed-form solution when only the mean and standard deviation of the online channel demand distribution is known and the offline channel demand follows a uniform distribution (partial robust). Specifically, owing to the good structural properties of the solution, we obtain a heuristic ordering formula for the general distribution case (i.e., the offline channel demand follows a general distribution). In addition, a series of numerical experiments prove the rationality of our conjecture. Moreover, after comparing our solution with other possible policies, we conclude that the PRO approach improves the performance of incorporating the internet into an existing supply chain and, thus, is able to adjust the level of conservativeness of the solution. Finally, in a degenerated situation, we compare our PRO approach with a combination of information approach. The results show that the PRO approach has more “robust” performance. As a result, a reasonable trade-off between robustness and performance is achieved.     

多个小波基的联合图像去噪方法

研究了利用多个小波基对含噪图像进行联合去噪的新方法。这种方法首先用每个小波基对含噪图像进行分解、阈值和恢复,得到多幅恢复图像;对这些图像进行简单的算术平均,平均图像的质量明显提高。与非下采样小波去噪方法相比,这种方法计算量小得多而性能相近。通过实验研究了小波基的选取问题,给出了一些经验原则。     

基于理想存取结构划分的(2, n)异或视觉密码

针对(2, n)异或视觉密码的完全恢复问题,依据理想存取结构定义,在分析最小授权子集之间特殊的共享份关系基础上,借助图论思想设计了一种存取结构划分算法,并证明了该算法划分得到的理想存取结构数目最少。设计并实现了秘密分享与恢复算法。实验结果表明,提出的(2, n)异或视觉密码方案可以实现秘密图像的完全恢复,且与现有方案相比,像素扩展度达到最优。     

Some new similarity measures for intuitionistic fuzzy values and their application in group decision making

We first propose a series of similarity measures for intuitionistic fuzzy values (IFVs) based on the intuitionistic fuzzy operators (Atanassov 1995). The parameters in the proposed similarity measures can control the degree of membership and the degree of non-membership of an IFV, which can reflect the decision maker’s risk preference. Moreover, we can obtain some known similarity measures when some fixed values are assigned to the parameters. Furthermore, we apply the similarity measures to aggregate IFVs and develop some aggregation operators, such as the intuitionistic fuzzy dependent averaging operator and the intuitionistic fuzzy dependent geometric operator, whose prominent characteristic is that the associated weights only depend on the aggregated intuitionistic fuzzy arguments and can relieve the influence of unfair arguments on the aggregated results. Based on these aggregation operators, we develop some group decision making methods, and finally extend our results to interval-valued intuitionistic fuzzy environment.     

相似产品步进应力加速寿命试验的统计分析

本文将单组样本的步进应力加速寿命试验(SSALT)拓展到双样本情形.基于累积失效模型(CEM),对两组指数产品的联合Ⅱ型截尾数据进行综合分析.首先,计算未知参数的极大似然估计(MLE),并利用矩母函数(MGF)建立其条件概率分布.然后,采用两种方法,基于概率分布的精确分析方法和参数的Bootstrap方法,构建参数的置信区间(CIs).以覆盖率作为评价标准,对两种方法的区间估计效果进行分析.最后,通过数值算例验证了本文方法的有效性.     

修理工多重休假且修理设备可更换的k/n(G)表决可修系统研究

本文将“修理工多重休假”机制和“修理设备可发生失效且可更换”策略同时引入到k/n(G)表决可修系统中, 利用马尔可夫过程理论求得了系统处于各状态的稳态概率的递推表达式. 在此基础上,给出了系统的稳态可用度、首次故障前平均时间、稳态故障频度、修理工繁忙的稳态概率、修理设备的稳态不可用度、故障部件的平均数目以及故障部件的平均等待修理时间等一系列性能指标. 最后以6/10(G)表决可修系统为例,分析了修理工的休假率和修理设备的失效率对几个主要性能指标的影响.