Implicit Bayesian Inference Using Option Prices

Abstract. A Bayesian approach to option pricing is presented in which posterior inference about the underlying returns process is conducted implicitly via observed option prices. A range of models allowing for conditional leptokurtosis, skewness and time‐varying volatility in returns are considered, with posterior parameter distributions and model probabilities backed out from the option prices. Models are ranked according to several criteria, including out‐of‐sample predictive and hedging performance. The methodology accommodates heteroscedasticity and autocorrelation in the option pricing errors, as well as regime shifts across contract groups. The method is applied to intraday option price data on the S&P500 stock index for 1995. While the results provide support for models that accommodate leptokurtosis and skewness, no one model dominates when all criteria are considered.     

基于改进Kendall-f算法的发电机定子温度分析

在线监测发电机定子温度可有效发现定子部分过热性故障,如何从检测的温度数据中提取反映定子故障预警的信息是目前国内外研究的热点。该文采用可变窗改进Kendall-f算法处理监测温度数据,提取其变化趋势特征信息;采用偏度分析提取监测温度异常值,从而正确诊断定子过热性故障,仿真分析及实验结果表明该方法有效提高了定子温度监测与预警的准确率。     

在癌症分类中基于分层抽样的神经网络集成算法

在基因表达谱数据的分析中,针对有效合理地选择特征基因集的问题,本文将分层抽样技术引入特征基因选择,提高特征基因集的分类能力。以神经网络作为分量分类器,神经网络集成进行分类预测。并在结肠癌数据集上进行实验,实验结果表明该方法能有效地降低特征基因集选择的复杂性,提高对于未知样本的分类预测效果。     

圆柱绕流尾涡图像的灰度特征和纹理特征

图像技术是流体力学研究的重要手段,在流动显示和流动测量中应用广泛;随着信息技术的发展,图像技术在流动规律研究的应用不断深入。论文以圆柱绕流为对象,分析雷诺数Re = 50 ~ 1725时尾涡图像的分布特征,解析图像的灰度特征参数和纹理特征参数随雷诺数的变化规律,并开展流动特性解释。研究发现,圆柱绕流的紊动扩散速率、分离点位置、尾涡空间分布的变化导致了图像灰度与纹理呈现出不同的识别特征：当Re < 600时,尾涡的灰度直方图的均值随雷诺数的增加线性递减;当Re = 148 ~ 642时,基于尾涡图像的纹理对比度推导出的尾涡脱落周期长度与经验取值吻合良好;当Re > 170时,灰度歪斜度随雷诺数递增且表现为明显的线性关系。     

On the performance of different capability indices under normal and non-normal distributions

In this article, we have examined the performance of some useful capability indices using normal and non-normal distributions. The confidence intervals are calculated and mean coverage rates are observed for different capability indices. The effects of symmetry and kurtosis of parent distributions are examined on the mean coverage rates of different capability indices. Moreover, we have investigated the robustness (of confidence interval) using the median and percentile-based indices. We have considered the well-known distributions including normal, gamma, t, Weibull, and chi-squared. For these process scenarios, we have observed that some indices resist disturbance only in symmetry of the parent distribution, some resist the disturbance in symmetry and kurtosis of the distribution, and some indices don’t resist against either type of disturbance.     

A skewed unscented Kalman filter

State estimation of nonlinear systems is a challenging task, especially when the Gaussian approximation fails. The unscented Kalman filter was proposed to deal with state estimation of nonlinear systems. We modify the traditional unscented Kalman filter to capture the third-order moment (skewness) of the state vector. Methods are also proposed to reduce the computation time of the suggested approach, and showing that the proposed algorithm is as fast as the unscented Kalman filter. Simulation results confirm that the method is better than, or at least as good as, the unscented Kalman filter.     

基于Hadoop的MapReduce模型的研究与改进

针对MapReduce模型中存在的多个Reduce任务之间完成时间差别较大的问题,分析了影响Reduce任务完成时间的因素,指出了MapReduce模型中Reduce任务节点存在数据倾斜问题,提出了一种改进型的MapReduce模型MBR(Map-Balance-Reduce)模型。通过添加Balance任务,对Map任务处理完成的中间数据进行均衡操作,使得分配到Reduce任务节点的数据比较均衡,从而确保Reduce任务的完成时间基本一致。仿真实验结果表明,经过Balance任务后,Map任务产生的中间数据能够比较均衡的分配给Reduce任务节点,达到数据计算均衡的目的,在一定程度上减少了整个作业的执行时间。     

基于高阶累积量的模拟电路故障诊断

结合信息融合技术,提出一种基于高阶累积量的模拟电路故障诊断方法.对待测电路施加正弦激励信号,从待测电路的输出终端提取原始电压和电流信号,求出峰度和斜度作为故障特征向量,将其输入改进的反向传播神经网络进行故障诊断.实例结果表明,该方法具有较高的诊断准确率.     

低信噪比条件下的调制方式识别

传统的调制方式识别方法通常假设接收信号是平稳随机过程,一般要在信噪比不低于10dB时才能取得较高的识别率。本文根据模拟调制信号的谱特性和谱相关特性,给出了低信噪比下的调制方式识别算法,理论分析和仿真结果表明,本文提出的算法,在低信噪比(3dB)条件下,仍然具有很高的识别率。     

The Identification of Data Distribution in Oil Production

Abstract

This article suggests nonparametric criteria for diagnosing changes in process condition based on analysis of data variation.

The developed criteria allow diagnosing changes in conditions of the studied process in cases of data distribution skewness, as well as cases when application of other criteria is not correct or complicated.

The suggested statistical criteria allow simplification of process diagnosis, which are described by multifractal, chaotic fluctuations, and their calculation algorithm can be easily realized.

The applicability of the diagnostic criteria has been proven both in modeled and practical examples of oil production.