基于Stacking集成学习的恒星/星系分类研究

机器学习在当今诸多领域已经取得了巨大的成功,但是机器学习的预测效果往往依赖于具体问题.集成学习通过综合多个基分类器来预测结果,因此,其适应各种场景的能力较强,分类准确率较高.基于斯隆数字巡天(Sloan Digital Sky Survey,SDSS)计划恒星/星系中最暗源星等集分类正确率低的问题,提出一种基于Stacking集成学习的恒星/星系分类算法.从SDSS-DR7(SDSS Data Release 7)中获取完整的测光数据集,并根据星等值划分为亮源星等集、暗源星等集和最暗源星等集.仅针对分类较为复杂且困难的最暗源星等集展开分类研究.首先,对最暗源星等集使用10折嵌套交叉验证,然后使用支持向量机(Support Vector Machine,SVM)、随机森林(Random Forest,RF)、XGBoost(eXtreme Gradient Boosting)等算法建立基分类器模型;使用梯度提升树(Gradient Boosting Decision Tree,GBDT)作为元分类器模型.最后,使用基于星系的分类正确率等指标,与功能树(Function Tree,FT)、SVM、RF、GBDT、XGBoost、堆叠降噪自编码(Stacked Denoising AutoEncoders,SDAE)、深度置信网络(Deep Belief Network,DBN)、深度感知决策树(Deep Perception Decision Tree,DPDT)等模型进行分类结果对比分析.实验结果表明,Stacking集成学习模型在最暗源星等集分类中要比FT算法的星系分类正确率提高了将近10%.同其他传统的机器学习算法、较强的提升算法、深度学习算法相比,Stacking集成学习模型也有较大的提升.     

A multi-equation spatial econometric model,with application to EU manufacturing productivity growth

A multi-equation spatial econometric model is used to explain variations across EU regions in manufacturing productivity growth based on recent theoretical developments in urban economics and economic geography. The paper shows that temporal and spatial parameter homogeneity is an unrealistic assumption, contrary to what is typically assumed in the literature. Constraints are imposed on parameters across time periods and between core and peripheral regions of the EU, with the significant loss of fit providing overwhelming evidence of parameter heterogeneity, although the final model does highlight increasing returns to scale, which is a central feature of contemporary theory.      

基于贝叶斯模型的器测、古气候重建与气候模拟数据的融合试验

气象部门馆藏的西部最早的器测气象资料始于20世纪30年代,不能满足建立20世纪以来中国气候变化序列的需求,而古气候重建或气候模拟资料则可以延伸到器测时代以前。为了探讨长序列多源气候资料序列融合方法,采用贝叶斯方法对中国北疆地区8条树轮气温重建资料、器测资料与国际耦合模式比较计划第5阶段（CMIP5）模式模拟资料进行了融合试验。首先利用器测资料对气温代用资料进行校验与网格重建,并以此作为贝叶斯模型的先验分布,然后,用泰勒图选出了该区域气候模拟效果最佳的几个模式;最后将网格重建和气候模拟序列用贝叶斯模型进行了融合试验。结果表明,贝叶斯融合模型能有效提取各种数据来源的有用信息进行融合,融合结果的长期变化（线性）趋势更接近器测气候序列,并在一定程度上提高了序列的精度,减小了结果的不确定性;并且,融合结果能够纠正先验分布及气候模拟数据的明显偏差,为长年代气候序列重建提供了一个可行的思路。      

IMPACT OF DIABATIC PROCESSES IN AGCM ON 4-DIMENSIONAL VARIATIONAL DATA ASSIMILATION*

The impact of diabatic processes on 4-dimensional variational data assimilation (4D-Var) was studied using the 1995 version of NCEP's global spectral model with and without full physics.The adjoint was coded manually.A cost function measuring spectral errors of 6-hour forecasts to "observation" (the NCEP reanalysis data) was minimized using the L-BFGS (the limited memory quasi-Newton algorithm developed by Broyden,Fletcher,Goldfard and Shanno) for optimizing parameters and initial conditions.Minimization of the cost function constrained by an adiabatic version of the NCEP global model converged to a minimum with a significant amount of decrease in the value of the cost function.Minimization of the cost function using the diabatic model, however,failed after a few iterations due to discontinuities introduced by physical parameterizations.Examination of the convergence of the cost function in different spectral domains reveals that the large-scale flow is adjusted during the first 10 iterations,in which discontinuous diabatic parameterizations play very little role.The adjustment produced by the minimization gradually moves to relatively smaller scales between 10-20th iterations.During this transition period,discontinuities in the cost function produced by "on-off" switches in the physical parameterizations caused the cost function to stay in a shallow local minimum instead of continuously decreasing toward a deeper minimum. Next,a mixed 4D-Var scheme is tested in which large-scale flows are first adiabatically adjusted to a sufficient level,followed by a diabatic adjustment introduced after 10 to 20 iterations. The mixed 4D-Var produced a closer fit of analysis to observations,with 38% and 41% more decrease in the values of the cost function and the norm of gradient,respectively,than the standard diabatic 4D-Var,while the CPU time is reduced by 21%.The resulting optimal initial conditions improve the short-range forecast skills of 48-hour statistics.The detrimental effect of parameterization discontinuities on minimization was also reduced.     

北京大灰厂试验场重力变化特征及其解释

介绍了北京大灰厂试验场（简称京西试验场）的概况，对１９９３年３月以来重力重复观测资料进行了处理。计算了各期重力观测精度，结果显示１９９３年３月～１９９５年５月京西试验场存在一定重力变化。采用给定的模型反演了八宝山断层介质的密度变化，认为断层破碎带介质密度的变化可能是引起地表重力变化的主要因素     

Developing constitutive models from EPR‐based self‐learning finite element analysis

A constitutive model that captures the material behavior under a wide range of loading conditions is essential for simulating complex boundary value problems. In recent years, some attempts have been made to develop constitutive models for finite element analysis using self‐learning simulation (SelfSim). Self‐learning simulation is an inverse analysis technique that extracts material behavior from some boundary measurements (eg, load and displacement). In the heart of the self‐learning framework is a neural network which is used to train and develop a constitutive model that represents the material behavior. It is generally known that neural networks suffer from a number of drawbacks. This paper utilizes evolutionary polynomial regression (EPR) in the framework of SelfSim within an automation process which is coded in Matlab environment. EPR is a hybrid data mining technique that uses a combination of a genetic algorithm and the least square method to search for mathematical equations to represent the behavior of a system. Two strategies of material modeling have been considered in the SelfSim‐based finite element analysis. These include a total stress‐strain strategy applied to analysis of a truss structure using synthetic measurement data and an incremental stress‐strain strategy applied to simulation of triaxial tests using experimental data. The results show that effective and accurate constitutive models can be developed from the proposed EPR‐based self‐learning finite element method. The EPR‐based self‐learning FEM can provide accurate predictions to engineering problems. The main advantages of using EPR over neural network are highlighted.     

基于变网格聚类的离群数据的研究

随着计算机技术的迅速发展,计算机在过程控制中起着越来越重要的作用,在集散控制系统中,上位机通过PCU(Process ControlUnit)传送来的数据进行分析处理而后对PCU实现参数整定、报警与管理。因此这一领域的数据挖掘方法研究具有十分重要的意义。文章将根据这一领域的特殊情况,提出一种基于变网格的聚类分析算法分析过程数据库中的离群数据,从而实现系统的预测报警功能。     

地面三维激光雷达点云误差分析及校正方法

分析了地面三维激光雷达点云误差来源,提出了相应的误差校正模型。通过实测数据验证,该模型能够消除系统误差,提高了点云精度。     

基于GIS技术的县域城市规划数据管理模式应用的研究——以广东省增城市为例

随着近年来GIS技术在城乡规划方面的应用日趋普及,GIS技术在空间数据采集、数据管理与分析等关键环节上的优势也逐渐显露。同时,县域城市城乡规划数据种类繁杂,数据量大,亟需先进的技术手段对其进行有效组织与管理。采用GIS技术对城乡规划数据进行动态管理和更新将给城乡规划带来极大的便利。本文以广东省增城市规划管理信息系统为案例,系统阐述GIS技术在城乡规划数据管理中的应用。     

基于马尔科夫链的北斗行驶记录仪数据存储算法研究

介绍了北斗行驶记录仪的组成和功能,对定时1 s和定时60 s报送或存储进行了对比和优劣分析,根据道路匹配特性,给出了一种基于马尔科夫链的北斗行驶记录仪数据存储算法,将马尔科夫链上的节点（拐点）作为记录仪存储或传输的位置点,通过采集车辆状态信息和定位模块输出信息对“拐点”计算方法进行分析和推导,并通过多次实际跑车测试,测试结果表明：与定时1 s报送或存储相比,节省率可达到90％以上,大大节省了存储空间或者通信费用,具有较好的使用效果。