Estimation of the Coefficient of Multiple Determination

Assume that we have iid observations on the random vector X = (X ,...,X ) following a multivariate normal distribution N (,) where both R and (p.d.) are unknown. Let denote the multiple correlation coefficient between X and (X ,...,X ). The parameter = , called the multiple coefficient of determination, indicates the proportion of variability in X explained by its best linear fit based on (X ,..., X ). In this paper we consider the point estimation of under the ordinary squared error loss function. The usual estimators (MLE, UMVUE) have complicated risk expressions and hence it is quite difficult to get exact decision-theoretic results. We therefore follow the asymptotic decision theoretic approach (as done by Ghosh and Sinha (1981, Ann. Statist., 9, 1334-1338)) and study Second Order Admissibility of various estimators including the usual ones.     

Deeply fusing multi-model quality-aware features for sophisticated human activity understanding

Detecting and understanding human action under sophisticated lighting condition and backgrounds, also known as human action recognition in real-world context, is an indispensable component in modern intelligent systems and has becoming a hot research topic currently. Nowadays, human action recognition is still a tough challenge due to intra-class and inter-class, environment and temporal-level differences of the same action. Algorithms based on the single visual channel cannot achieve satisfactory performance. Thus, in this paper, we propose a novel action recognition framework towards sophisticated activity understanding, focusing on intelligently combining multimodel quality-related action features. Specifically, we first design a multi-channel feature fusion (MCFF) algorithm to capture visual appearance, motion and acoustic patterns from each video frame, where image-level labels are characterized by choosing high quality multimodel features. Subsequently, we design an adaptive key frame selection algorithm that can be applied to characterize human action from human action video stream. Thereafter, we engineer a multimodel feature based on an auxiliary human action retrieval system to achieve sophisticated activity understanding. Extensive experimental evaluations have demonstrated that the effectiveness and robustness of our proposed method.     

基于回归分析的迎峰度夏用电量分析

利用回归分析算法,提出了一种迎峰度夏期间最高气温与日用电量关系分析方法,首先利用皮尔森相关系数对度夏期间的最高气温和日用电量相关程度进行定量分析,然后利用回归分析算法建立最高气温和日用电量的拟合函数,并利用最小二乘法进行求解,得到最高气温和日用电量相关关系的定量分析结果。研究结果表明,提出的度夏期间用电量分析方法能够定量地给出气温变化导致的用电量变化情况,并能根据分析结果对未来几天的日用电量进行准确预测。     

基于零相关区序列集的空时自适应处理地面运动目标检测算法

在多发单收的合成孔径雷达-地面运动目标检测(SAR鄄GMTI)雷达系统中,虽然多颗卫星之间不需要数据传送就能在接收卫星上实现地面运动目标检测,但存在由于空间发射互相正交的一组波形而导致SAR 成像相干积累的积分旁瓣过高的问题,从而影响对后续的图像域杂波相消性能。因此,文中提出了将空时自适应处理算法用于发射零相关区序列集(ZCZcodes)信号的星载多入单出系统,这样不用考虑积分旁瓣对SAR 成像质量和图像域杂波相消性能带来的影响,且计算量较SAR鄄GMTI 算法小,同样不需要数据传送也在接收卫星上实现了地面运动目标检测。仿真结果和性能分析验证了该方法的有效性。     

基于粒子群算法分析随机相位对全息图的质量影响

因为相息图具有很好的衍射效率,所以经常被用 在计算全息中。在计算相息图时,随机相位被加入 到物体对象中从而扩散全息图中的物体光,但随机相位的添加会造成重建图像中产生相当大 的散斑噪声。 为了改善这一问题,本文首先针对随机相位的范围进行分析,研究了初始的随机相位对全息 图成像质量的 影响,再通过粒子群算法寻找最优的随机相位范围。在计算全息图时会存在部分信息的丢失 ,所以对全息 图进行迭代运算能弥补全息图间的差异。故在本文还将结合改进的迭代傅里叶变换算法来有 效的改善重建 图像的质量,即提出了一种基于粒子群算法分析有限随机相位对全息图质量影响的方法。通 过对以上过程 进行数值模拟和光学实验,最后的实验结果表明,通过寻找最优的随机相位范围以及改进的 迭代傅里叶变 换算法能够改善重建图像的散斑噪声得到高质量的重建像。     

基于多模激光关联光谱的甲烷气体检测技术

为提高多模激光吸收光谱的灵敏度,将多模二极 管激光关联光谱、长程吸收技术和波长调制光谱结合,建立了一 套具有高检测灵敏度和高稳定性的气体检测系统。采用1675nm多模激光器作为光源,以光程为100 m的离散镜片型 多通 池作为气体吸收池,利用甲烷的2ν₃泛频带吸收谱线,通过计 算待测气体和参考气体的二次谐波信号峰值之间的关系,实 现了对CH₄气体的测量。实验在室温和20.265 kPa(即0.2个标准大气压)条件下进行 ,CH₄浓度范围为1.00×10－5－1.10×10－2。实验结果表明,CH₄浓度测量值与真实值之间具有良好的线性关系, 其线性度为0.997,系统的测量准确度为 3.50%,对浓度为2.60×10－6的气体样 品在30 min内的连续测 量表明系统稳定度优于2.53%。本方法具有操作简单、稳定性 好、灵敏度高和环境适应性强等优点,在工业过程控制和环境监测中具有广阔的应用前景。     

Direct View on Non-Equilibrium Heterogeneous Dynamics in Glassy Nanorods

Amorphous solids are potential candidates for room temperature applications, such as structural materials, due to their outstanding mechanical properties. For the first time, the local atomic dynamics of amorphous metallic nanorods are spatially resolved at room temperature using electron microscopy. Methodologically, quasi-equilibrium conditions are verified by two-time correlation functions. FeNiP is chosen as a suitable model system since a bamboo structure forms upon electrodeposition into nanotubular confinement, consisting of thin Fe-rich and Ni-rich amorphous layers. Therefore, the present nano-glassy structure allows studying the glass dynamics of two glassy phases at once, at the same time avoiding any preparation-related structural artifacts of the intrinsically electron-transparent samples. By comparing the local atomic mobility with the local composition and the local structural information of the glassy phases, correlations between medium-range order and time scales of heterogeneous glassy dynamics are consistently obtained.     

Inter-layer correlation-based adaptive bit allocation for enhancement layer in scalable high efficiency video coding

To facilitate the diversity of network and end-user devices, bit allocation technology must be combined with scalable high efficiency video coding (SHVC) to achieve continuous bitrate variation. However, to date, the multilayer coding feature in SHVC has not been fully utilized. In this paper, a bit allocation algorithm is proposed for the enhancement layer in SHVC. The algorithm includes two parts: a bit allocation method for the initial frame and another for subsequent frames. The bit allocation method for the initial frame consists of a bit allocation factor model constructed based on investigation of the influence of the initial target bitrate on the overall coding performance. For the bit allocation of the subsequent frames, alternate GOP coding is designed and implemented to fully exploit the inter layer correlation. In addition, an adaptive frame layer bit allocation ratio model is deduced according to the rate distortion optimization theory. Experimental results show that the proposed algorithm can improve PSNR by 0.44 dB and 0.41 dB under low delay and random-access configurations, respectively, and achieve high bitrate control accuracy. The algorithm also outperforms the state-of-the-art algorithms in term of PSNR improvement.     

EEMD法在激光语音传输系统中的应用

为消除激光语音传输系统中激光器输出功率不稳定对探测信号的影响,提出了一种结合集合经验模态分解(EEMD)与相关系数的方法对信号进行处理.该方法将信号分解为不同的分量,算出不同分量和原信号间的相关系数,再设置固定判定阈值来区分真实信号和趋势项分量.该方法可以克服传统处理方法中存在的模态混叠和需要主观判断趋势分量的缺点,从而达到避免主观判断失误和准确提取趋势的目的.仿真及实验结果表明,使用该方法后再去噪效果优于直接去噪,此外该方法能有效去除激光器功率线性变化和频率小于15 Hz的正弦起伏变化导致的趋势.     

TD-LTE网内干扰识别方法研究

本文针对TD-LTE网内干扰识别的难点问题,分析了目前IDS系统的TD-LTE网内干扰识别算法的准确率,然后选取并对比分析了同频邻区不同业务量指标与高干扰小区24小时底噪均值的相关系数,确定了与底噪相关性最强的业务量指标。在此基础上,本文提出了一种基于同频邻区业务量与底噪相关性的网内干扰识别算法,大幅提升了TD-LTE网内干扰识别准确率。