基于贝叶斯网络的移动支付风险评估模型

随着信息技术和网络的迅猛发展,支付业务、技术及工具不断创新,移动支付的发展在逐渐加快。移动支付给人们生活带来方便和快捷的同时,也存在着较高的潜在风险,容易遭受非法入侵和恶意攻击。就移动支付风险的分析及风险值的计算理论方面开展工作,在贝叶斯网络的基础上,针对移动支付的主要组成主体,提出移动支付风险评估模型,通过使用该模型进行移动支付风险评估不仅可以对目前移动支付的风险进行评估,还可以根据风险评估结果引导风险控制,对比风险控制前后的风险值判断风险控制的效果,通过案例分析,提出的移动支付风险评估模型可以很好地完成移动支付的风险评估要求。     

基于需求更新的救灾品配送公平与效率协调模型

鉴于需求准确度与救灾品配送效率之间的悖反关系,研究了基于需求更新的救灾品配送效率与公平问题.通过对某市民政局进行调研,了解到台风灾害发生时的需求变化规律,设计一种通过已知需求灾区对未知需求灾区的需求进行贝叶斯更新的需求更新方式,由此建立救灾品配送模型.将该模型应用于某台风救灾实例,通过实例分析表明了该模型的求解速率与精度,并找到了最优"决策点",实现了救灾品配送公平与效率的平衡.     

基于WSN的精准养牛业智能决策支持系统设计

准确及时地获取牛群的身体变化情况是保证牛群生存环境舒适度的基本前提。在对养牛业监测系统和无线传感网络深入了解的情况下,开发了以无线传感网络为基础的精准养牛业系统。按照标准生产化的要求,系统的总体架构分为感知层、传输层和应用层。并根据总体架构的层次模型设计了相应的软件和硬件。系统有效地利用贝叶斯网络智能决策支持系统建立模型,结合无线传感网络收集牛酮体的数据,进行相应的环境设置,让牛群的生存环境更加舒适。     

Multimodel analysis and controller design for nonlinear processes

Multimodel analysis and controller design for nonlinear processes via gap metric is discussed. It is shown that the loop-shaping H_∞ approach can integrate the procedure of selecting operating points and the local controller design. The local controllers can guarantee not only stability but also performance specified by the pre- and/or post-compensators. Thus, at each operating points, local controllers can have similar performance, and the global performance of the system can be predicted.     

一种结合主题模型的推荐算法

针对传统协同过滤推荐算法存在的冷启动、数据稀疏以及相似度度量的准确性问题,基于LDA主题模型对文本隐式主题挖掘的有效性和KL散度在主题分布相似性度量的准确性,提出了结合LDA主题模型的矩阵分解推荐算法。首先,利用改进的LDA算法输出项目-主题分布,并用困惑度作为主题数设置的修正函数;然后分别基于余弦相似度和KL散度计算得到项目相似度矩阵,将得到的相似度矩阵结合原评分训练集输出预评分,再将预评分填充到训练集;最后将训练集输入ALS矩阵分解算法得到推荐结果。通过MovieLens数据集的实验结果表明,该算法在不同隐式参数设定下均能得到比ALS推荐算法以及更小的预测误差,并且最优预测误差小于传统推荐算法。该实验说明了通过集成LDA主题模型的ALS算法效果要优于其他推荐算法。     

Bayesian estimation of Dirichlet mixture model with variational inference

In statistical modeling, parameter estimation is an essential and challengeable task. Estimation of the parameters in the Dirichlet mixture model (DMM) is analytically intractable, due to the integral expressions of the gamma function and its corresponding derivatives. We introduce a Bayesian estimation strategy to estimate the posterior distribution of the parameters in DMM. By assuming the gamma distribution as the prior to each parameter, we approximate both the prior and the posterior distribution of the parameters with a product of several mutually independent gamma distributions. The extended factorized approximation method is applied to introduce a single lower-bound to the variational objective function and an analytically tractable estimation solution is derived. Moreover, there is only one function that is maximized during iterations and, therefore, the convergence of the proposed algorithm is theoretically guaranteed. With synthesized data, the proposed method shows the advantages over the EM-based method and the previously proposed Bayesian estimation method. With two important multimedia signal processing applications, the good performance of the proposed Bayesian estimation method is demonstrated.     

Coordinating efficiency and equity in disaster relief logistics via information updates

This paper addresses a multi-supplier, multi-affected area, multi-relief, and multi-vehicle relief allocation problem in disaster relief logistics. A multi-objective optimisation model based on disaster scenario information updates is proposed in an attempt to coordinate efficiency and equity through timely and appropriate decisions regarding issues such as vehicle routing and relief allocation. An optimal stopping rule is also proposed to determine the optimum period of delay before responding to disaster, because decision making requires accurate disaster information. The main contribution of this paper is solving relief allocation problem in a novel way by correlating operational research with statistical decision making and Bayesian sequential analysis. Finally, a case is presented based on the post-disaster rescue in Eastern China after supertyphoon Saomai to test the applicability and show the potential advantages of the proposed model.     

Towards subject independent continuous sign language recognition: A segment and merge approach

This paper presents a segment-based probabilistic approach to robustly recognize continuous sign language sentences. The recognition strategy is based on a two-layer conditional random field (CRF) model, where the lower layer processes the component channels and provides outputs to the upper layer for sign recognition. The continuously signed sentences are first segmented, and the sub-segments are labeled SIGN or ME (movement epenthesis) by a Bayesian network (BN) which fuses the outputs of independent CRF and support vector machine (SVM) classifiers. The sub-segments labeled as ME are discarded and the remaining SIGN sub-segments are merged and recognized by the two-layer CRF classifier; for this we have proposed a new algorithm based on the semi-Markov CRF decoding scheme. With eight signers, we obtained a recall rate of 95.7% and a precision of 96.6% for unseen samples from seen signers, and a recall rate of 86.6% and a precision of 89.9% for unseen signers.     

基于k-最近邻筛选的BMA集合预报模型研究

针对冗余训练样本会降低BMA参数求解效率与精度问题,本文提出在BMA运算之前采用k-最近邻(k-nearest neighbor)算法筛选有价值训练样本,并用于BMA参数求解的改进模型。模拟试验在淮河王家坝站进行,分别以k-最近邻筛选、不筛选两种方案为BMA提供训练样本,统计分析两种方案中王家坝站流量模拟结果,评价BMA改进法的性能。模拟结果显示,采用k-最近邻样本筛选方法后,BMA模型对洪水过程以及洪峰的预报精度提升明显;概率预报结果的离散程度降低的同时,可靠性程度获得提升。k-最近邻样本筛选方法的引入,能够有效去除BMA模型训练样本中的冗余数据,以少量的样本获得更可靠的模型参数,改善集合预报性能。     

Exploring approaches to improve the performance of autonomous monitoring with imperfect data in location-aware wireless sensor networks

In recent years, information and sensing technologies have been applied to the construction industry to collect and provide rich information to facilitate decision making processes. One of the applications is using location data to support autonomous crane safety monitoring (e.g., collision avoidance and dangerous areas control). Several location-aware wireless technologies such as GPS (Global Positioning System), RFID (Radio-frequency identification), and Ultra-Wide Band sensors, have been proposed to provide location information for autonomous safety monitoring. However, previous studies indicated that imperfections (errors, uncertainty, and inconsistency) exist in the data collected from those sensors and the data imperfections have great impacts on autonomous safety monitoring system performance. This paper explores five computationally light-weight approaches to deal with the data imperfections, aiming to improve the system performance. The authors built a scaled autonomous crane safety monitoring testbed with a mounted localization system to collect location data and developed five representative test cases based on a live construction jobsite. Seven hundred and sixty location readings were collected at thirty-eight test points from the sensors. Those location data was fed into the reasoning mechanisms with five approaches to generate the safety decisions at those thirty-eight test points and evaluate system performance in terms of precision, recall and accuracy. The results indicate that system performance can be improved if at least ten position readings from sensors can be collected at small intervals at any location along the moving path. However, by including additional data such as velocity and acceleration that may be read from devices mounted on workers, localization error may be significantly reduced. These findings represent a path forward to improve localization accuracy by mixing imperfect data from the sensed environment with supplemental input.