Topological properties of phylogenetic trees in evolutionary models

The extent to which evolutionary processes affect the shape of phylogenetic trees is an important open question. Analyses of small trees seem to detect non-trivial asymmetries which are usually ascribed to the presence of correlations in speciation rates. Many models used to construct phylogenetic trees have an algorithmic nature and are rarely biologically grounded. In this article, we analyze the topological properties of phylogenetic trees generated by different evolutionary models (populations of RNA sequences and a simple model with inheritance and mutation) and compare them with the trees produced by known uncorrelated models as the backward coalescent, paying special attention to large trees. Our results demonstrate that evolutionary parameters as mutation rate or selection pressure have a weak influence on the scaling behavior of the trees, while the size of phylogenies strongly affects measured scaling exponents. Within statistical errors, the topological properties of phylogenies generated by evolutionary models are compatible with those measured in balanced, uncorrelated trees.     

基于神经网络模糊控制理论的转台伺服系统控制设计

针对三轴飞行仿真转台伺服系统非线性、模型不精确等特点,在分析转台系统结构的基础上,采用模糊PID控制的方法对转台伺服系统进行仿真控制,得到较好的控制效果。模糊控制控制规则的获得带有很大的人为因素,并且在控制过程中对规则采用查表法占用大量的内存。基于以上原因,设计了神经网络模糊控制器(NNSOC),利用神经网络控制自学习、自调整的能力,为模糊控制器提供自动生成控制规则的能力;同时由于神经网络具有联系记忆能力,可对未训练的样本做出决策。对NNSOC的控制效果进行了仿真。结果表明：其具有很好的动态性能和鲁棒性,对转台的控制效果良好。     

Characterization of spherical particles using high-order neural networks and scanning flow cytometry

We retrieve the radius R, real n and imaginary k parts of the refractive index of homogeneous spherical particles using angular distribution of the light-scattering intensity. To solve the inverse light-scattering problem we use a high-order neural-network technique. The effect of network parameters on optimization is examined. The technique is evaluated for noise-corrupted input data at 0.6 μm<R<10.6 μm, 1.02<n<1.38, and 0<k<0.03. The errors of retrieval for nonabsorbing particles do not exceed 0.05 μm for radius and 0.015 for refractive index. The experimental verification is fulfilled by experimental data retrieved by means of a scanning flow cytometer. The light-scattering profiles of polystyrene beads and spherized red blood cells are processed with the high-order neural networks and a non-linear regression at Mie theory. The parameters retrieved by the high-order neural networks correlate well with the parameters retrieved by the least-square method.     

Inevitable self-similar topology of binary trees and their diverse hierarchical density

Self-similar topology, which can be characterized as power law size distribution, has been found in diverse tree networks ranging from river networks to taxonomic trees. In this study, we find that the statistical self-similar topology is an inevitable consequence of any full binary tree organization. We show this by coding a binary tree as a unique bifurcation string. This coding scheme allows us to investigate trees over the realm from deterministic to entirely random trees. To obtain partial random trees, partial random perturbation is added to the deterministic trees by an operator similar to that used in genetic algorithms. Our analysis shows that the hierarchical density of binary trees is more diverse than has been described in earlier studies. We find that the connectivity structure of river networks is far from strict self-similar trees. On the other hand, organization of some social networks is close to deterministic supercritical trees.     

时频域分形维数分析的光谱信号重叠峰解析算法

由于光谱谱线存在自然展宽、多普勒展宽、碰撞展宽等,使混合气体中多种成分的吸收光谱信号出现相邻谱峰重叠现象,给混合气体组成成分的定性或定量检测带来较大的困难。现有的方法在获取先验知识、处理精度、运算效率等方面存在不足。提出基于时频域分形维数分析的光谱信号重叠峰解析算法,结合小波的多尺度观测能力和分形的自相似度的度量能力,识别、定位和解析光谱信号中的重叠峰。首先利用小波对具有重叠谱峰的光谱信号进行光谱频率域和尺度域的分析,然后对该时频域的光谱信号在同一光谱频率下的多尺度数据进行自相似性度量和分形计算。逐频率计算后得到光谱信号在频率域的分形维数曲线。该曲线体现了光谱信号在不同尺度的自相似性,其极值位置与光谱信号的各独立峰的位置具有相关性。依据此特性,结合分形曲线的特征参数,最后利用神经网络解析出对应混合气体成分的混叠在一起的各个独立谱峰。该方法利用小波的多分辨率特性,对信号进行不同尺度的精细度量。分形模型则提高了系统解析复杂信号的能力,对重叠程度高的多谱峰重叠信号也有很强的处理能力。借助人工神经网络,实现了整个算法的自动测量。通过实验结果分析,验证了算法的有效性,并讨论影响算法效果的主要因素。     

红外光谱统计波长选择方法用于废气检测

为了从含有噪声的大气红外光谱中提取微量待测污染组分的定量特征,进而建立校正模型,本文提出了一种基于统计理论的波长选择方法。该方法针对待测组分,在对光谱各波长位置的噪声强度进行统计估计的基础上,提出了选择最佳波长子集的目标函数。该目标函数包含波长子集的噪声和长度参数,这使得在最小化模型误差的同时也防止了模型规模的无限制膨胀。为了检验该方法的性能。文章利用含背景噪声的实测光谱数据针对三种气体进行了波长选择,并利用神经网络技术分别建立了校正模型。实验结果与波长子集的优化结果相符,所选择的波长子集的长度不足光谱波长总点数的2％,同时,光谱中的噪声也在校正模型中得到了明显的抑制。实验结果证明了该波长选择方法的有效性。     

Simultaneous Determination of Multicomponents in Air Toxic Organic Compounds Using Artificial Neural Networks in Ftir Spectroscopy

The application of Artificial Neural Networks (ANNs) for nonlinear multivariate calibration using simulated FTIR data was demonstrated in this paper. Neural networks consisting of three layers of nodes were trained by using the back-propagation learning rule. Since parameters affect the performance of the network greatly, simulated data were used to train the network in order to get a satisfactory combination of all parameters. The mixtures of four air toxic organic compounds whose FTIR spectra are overlapped were chosen to evaluate the calibration and prediction ability of the network. The relative standard error (RSD%), the percent standard error of prediction samples (%SEP) and the percent standard error of calibration samples (%SEC) are used for evaluating the ability of the neural network.     

结合区域提取和改进卷积神经网络的水下小目标检测

针对水下小目标信息量有限而难以提取有效特征导致的检测性能不佳问题,提出了一种结合区域提取和融合Hu矩特征的改进卷积神经网络水下小目标检测方法。该方法包含区域提取和分类两个步骤。首先以马尔可夫随机场分割算法为基础进行区域提取,对潜在目标定位的同时降低伪目标对后续分类的干扰;然后提取潜在目标区域的Hu矩特征并融入卷积神经网络,形成一种形状特征表征能力更强的改进卷积神经网络用于分类。声呐实测数据处理结果表明,该方法可以有效提升对水下小目标的发现概率和正确报警率,与其他目标检测方法相比,该方法具有更好的检测性能和泛化性。     

Relaxation techniques for strictly convex network problems

Gauss—Seidel type relaxation techniques are applied in the context of strictly convex pure networks with separable cost functions. The algorithm is an extension of the Bertsekas—Tseng approach for solving the linear network problem and its dual as a pair of monotropic programming problems. The method is extended to cover the class of generalized network problems. Alternative internal tactics for the dual problem are examined. Computational experiments — aimed at the improved efficiency of the algorithm — are presented.This research was supported in part by National Science Foundation Grant No. DCR-8401098-A01.     

On a problem of Hornik

In 1991,Hornik proved that the collection of single hidden layer feedforward neural networks(SLFNs)with continuous,bounded,and non-constant activation functionσis dense in C(K)where K is a compact set in R~s(see Neural Networks,4(2),251-257(1991)).Meanwhile,he pointed out"Whether or not the continuity assumption can entirely be dropped is still an open quite challenging problem".This paper replies in the affirmative to the problem and proves that for bounded and continuous almost everywhere(a.e.)activation functionσon R,the collection of SLFNs is dense in C(K)if and only ifσis un-constant a.e..