Local spatial properties based image interpolation scheme using SVMs

Image interpolation plays an important role in image process applications. A novel support vector machines （SVMs） based interpolation scheme is proposed with increasing the local spatial properties in the source image as SVMs input patterns. After the proper neighbor pixels region is selected, trained support vectors are obtained by training SVMs with local spatial properties that include the average of the neighbor pixels gray values and the gray value variations between neighbor pixels in the selected region. The support vector regression machines are employed to estimate the gray values of unknown pixels with the neighbor pixels and local spatial properties information. Some interpolation experiments show that the proposed scheme is superior to the linear, cubic, neural network and other SVMs based interpolation approaches.     

基于灰色组合模型的河南省粮食产量预测

一元线性回归有直线趋势,而GM(1,1)能较好地模拟指数变化的趋势。但是,如果原始序列整体上是直线趋势,在少数点上,数据模拟值和回归直线偏离较大时,线性函数已不能很好地预测数据序列的变化了。对于此类问题,将数据分为跳变点(即模拟值偏离回归直线较大)和非跳变点数据,并将跳变点又分为上、下跳变点,借鉴灰色灾变预测原理,用GM(1,1)模型预测跳变点数据,而对其他非跳变点使用去掉跳变点后的数据形成的新的线性回归方程进行预测。通过对河南省粮食产量的预测,结果表明该方法很好地克服了GM(1,1)模型和线性回归模型的缺陷,具有较好的实际应用价值。     

基于GA优化小波LS-SVR的惯性器件故障预报

为了提高最小二乘支持向量回归机的性能,将Morlet小波核函数引入其中,形成了最小二乘小波支持向量回归机模型。利用待优化的参数重构模型的目标函数和约束条件,并在此基础上通过遗传算法进行参数选择,从而提高了该模型的泛化能力。将最小二乘小波支持向量回归机应用于导弹陀螺仪的漂移趋势预测,仿真实验结果表明了该方法的有效性和可行性,因此可以为陀螺仪的故障预报、可靠性辅助决策提供依据。     

Pause or No Pause？--Prosodic Phrase Boundaries Revisited

This study presents evidence from analyses of the acoustic parameters of fluent continuous speech to show that within-paragraph prosodic phrase boundaries are related more to contrasts of neighborhood prosodic states rather than between-phrase pause durations; prosodic states receive more constraints from higher level discourse information. By revising a modular acoustic model by Tseng＇s hierarchical prosodic phrase grouping framework and examining the much varied prosodic phrase （PPh） boundary B3 within speech paragraph, we show that statistical accounts of layered contributions reveal distinct contrasts between boundary immediate duration and intensity patterns irrespective of pause duration. Contrasts of FO contour patterns were also observed in these locations. Evidence was also obtained to illustrate how PPh boundary states are specified more by higher level discourse information than by lower level prosodic word construction. These combined results suggest that contrastive neighboring prosodic states are more significant cues to PPh boundaries than boundary pause duration. The results also help explain why in fluent speech between-phrase pause durations vary greatly, and can be applied to automatic speech segmentation.     

Pause or No Pause?——Prosodic Phrase Boundaries Revisited

This study presents evidence from analyses of the acoustic parameters of fluent continuous speech to show that within-paragraph prosodic phrase boundaries are related more to contrasts of neighborhood prosodic states rather than between-phrase pause durations; prosodic states receive more constraints from higher level discourse information.By revising a modular acoustic model by Tseng's hier-archical prosodic phrase grouping framework and examining the much varied prosodic phrase (PPh) boundary B3 within speech paragraph,we show that statistical accounts of layered contributions reveal dis-tinct contrasts between boundary immediate duration and intensity patterns irrespective of pause duration.Contrasts of F0 contour patterns were also observed in these locations.Evidence was also obtained to illus-trate how PPh boundary states are specified more by higher level discourse information than by lower level prosodic word construction.These combined results suggest that contrastive neighboring prosodic states are more significant cues to PPh boundaries than boundary pause duration.The results also help explain why in fluent speech between-phrase pause durations vary greatly,and can be applied to automatic speech segmentation.     

四川南充市农业劳动力状况统计分析与预测

根据南充农村劳动近几年的分布情况，运用多元线性回归分析，偏相关系数等统计方法，找出了影响南充农业劳动力分布情况的主要因素及其规律，并对未来十年南充在市人口及从事农业人口进行了预测与比较，指出南充粮食可以自给，城镇化仍在继续．     

用Cp统计量选择多元线性模型及估计形式

本文根据风险函数,提出了一个多变量场合的Cp统计量作为风险的估计,按Cp最小的原则给出了一种选择多元线性模型及估计形式的方法,重点讨论多变量正交试验的情况。     

售电量的模糊预测

售电量中,长期预测是电力企业经济分析的重要环节,是计划、决策的基础。 影响售电量的因素很多,也很复杂,如:工农业总产值、城市人口,家用电器的饱和程度,售电政策等。 本文把售电量及其影响因素视作模糊系统,用模糊回归修正模型预测售电量,通过相对误差检验,精度很高。     

非参数回归的CV NN中位数估计的相合性

考虑非参数回归模型Y_i=g(x_i)+e_i,其中g(x)是待估的连续函数,x_i是非随机的,e_i是i.i.d.随机误差。笔者讨论最近邻中位数估计g_(n,h)(x_i)=m(Y_i(1),…,Y_i(h))=Y_i(1),…,Y_i(h)的中位数,其中h利用平均平方误差意义下的cross-validation方法选择。在一定条件下,建立了cross-validation最近邻中位数估计的相合性。