Identification of equivalent dynamics using ordinal pattern distributions

The frequency of occurrence of ordinal patterns in an observed (or measured) times series can be used to identify equivalent dynamical system. We demonstrate this approach for system identification and parameter estimation for dynamics that can (at least approximately) be described by one-dimensional iterated maps.     

Tremor classification and tremor time series analysis

The separation between physiologic tremor (PT) in normal subjects and the pathological tremors of essential tremor (ET) or Parkinson's disease (PD) was investigated on the basis of monoaxial accelerometric recordings of 35 s hand tremor epochs. Frequency and amplitude were insufficient to separate between these conditions, except for the trivial distinction between normal and pathologic tremors that is already defined on the basis of amplitude. We found that waveform analysis revealed highly significant differences between normal and pathologic tremors, and, more importantly, among different forms of pathologic tremors. We found in our group of 25 patients with PT and 15 with ET a reasonable distinction with the third momentum and the time reversal invariance. A nearly complete distinction between these two conditions on the basis of the asymmetric decay of the autocorrelation function. We conclude that time series analysis can probably be developed into a powerful tool for the objective analysis of tremors. (c) 1995 American Institute of Physics.     

Cross and joint ordinal partition transition networks for multivariate time series analysis

We propose the construction of cross and joint ordinal pattern transition networks from multivariate time series for two coupled systems, where synchronizations are often present. In particular, we focus on phase synchronization, which is a prototypical scenario in dynamical systems. We systematically show that cross and joint ordinal pattern transition networks are sensitive to phase synchronization. Furthermore, we find that some particular missing ordinal patterns play crucial roles in forming the detailed structures in the parameter space, whereas the calculations of permutation entropy measures often do not. We conclude that cross and joint ordinal partition transition network approaches provide complementary insights into the traditional symbolic analysis of synchronization transitions.     

Multivariate analysis of fMRI time series: classification and regression of brain responses using machine learning

Machine learning and pattern recognition techniques are being increasingly employed in functional magnetic resonance imaging (fMRI) data analysis. By taking into account the full spatial pattern of brain activity measured simultaneously at many locations, these methods allow detecting subtle, non-strictly localized effects that may remain invisible to the conventional analysis with univariate statistical methods. In typical fMRI applications, pattern recognition algorithms "learn" a functional relationship between brain response patterns and a perceptual, cognitive or behavioral state of a subject expressed in terms of a label, which may assume discrete (classification) or continuous (regression) values. This learned functional relationship is then used to predict the unseen labels from a new data set ("brain reading"). In this article, we describe the mathematical foundations of machine learning applications in fMRI. We focus on two methods, support vector machines and relevance vector machines, which are respectively suited for the classification and regression of fMRI patterns. Furthermore, by means of several examples and applications, we illustrate and discuss the methodological challenges of using machine learning algorithms in the context of fMRI data analysis.     

A new complexity measure for time series analysis and classification

Complexity measures are used in a number of applications including extraction of information from data such as ecological time series, detection of non-random structure in biomedical signals, testing of random number generators, language recognition and authorship attribution etc. Different complexity measures proposed in the literature like Shannon entropy, Relative entropy, Lempel-Ziv, Kolmogrov and Algorithmic complexity are mostly ineffective in analyzing short sequences that are further corrupted with noise. To address this problem, we propose a new complexity measure ETC and define it as the “Effort To Compress” the input sequence by a lossless compression algorithm. Here, we employ the lossless compression algorithm known as Non-Sequential Recursive Pair Substitution (NSRPS) and define ETC as the number of iterations needed for NSRPS to transform the input sequence to a constant sequence. We demonstrate the utility of ETC in two applications. ETC is shown to have better correlation with Lyapunov exponent than Shannon entropy even with relatively short and noisy time series. The measure also has a greater rate of success in automatic identification and classification of short noisy sequences, compared to entropy and a popular measure based on Lempel-Ziv compression (implemented by Gzip).     

A coherence-based approach for the pattern recognition of time series

A pattern recognition approach based on the frequency domain measure of squared coherence is a useful approach to identify linearly related groupings of time series over different periods of time. It is considered in an application to identify similar patterns of the yearly rates of change in the Gross Domestic Product (GDP) of twenty two highly developed countries in an econophysics context. The approach is also tested in simulation studies using linearly related time series, and it is shown to have a very good success rate of correct pattern matching.     

Ordinal pattern and statistical complexity analysis of daily stream flow time series

When calculating the Bandt and Pompe ordinal pattern distribution from given time series at depth D, some of the D! patterns might not appear. This could be a pure finite size effect (missing patterns) or due to dynamical properties of the observed system (forbidden patterns). For pure noise, no forbidden patterns occur, contrary to deterministic chaotic maps. We investigate long time series of river runoff for missing patterns and calculate two global properties of their pattern distributions: the Permutation Entropy and the Permutation Statistical Complexity. This is compared to purely stochastic but long-range correlated processes, the k-noise (noise with power spectrum f ^?k ), where k is a parameter determining the strength of the correlations. Although these processes closely resemble runoff series in their correlation behavior, the ordinal pattern statistics reveals qualitative differences, which can be phrased in terms of missing patterns behavior or the temporal asymmetry of the observed series. For the latter, an index is developed in the paper, which may be used to quantify the asymmetry of natural processes as opposed to artificially generated data.     

一种串联组合分类的红外行人检测方法

针对红外图像行人检测中虚警率高的问题,提出了一种串联组合分类的红外行人检测方法。该方法利用感兴趣区域的宽高比和方向梯度直方图2种特征,采用串联组合方式实现分类识别。实验结果中检测率达到了90.5%,虚警率减少了43.7%,表明提出的方法可有效检测行人同时降低虚警。     

Quantifying postural sway dynamics using burstiness and interevent time distributions

The European Physical Journal B - We present numerical evidence using Monte Carlo simulations of finite temperature phase transition in two dimensional Coulomb Glass lattice model with random site...     

基于时间序列符号化模式表征的有向加权复杂网络

时间序列复杂网络分析近些年已发展成为非线性信号分析领域的一个国际热点课题.为了能更有效地挖掘时间序列(特别是非线性时间序列)中的结构特征,同时简化时间序列分析的复杂度,提出了一种新的基于时间序列符号化结合滑窗技术模式表征的有向加权复杂网络建网方法.该方法首先按照等概率区段划分的方式将时间序列做符号化处理,结合滑窗技术确定不同时刻的符号化模式作为网络的节点;然后将待分析时间序列符号化模式的转换频次和方向作为网络连边的权重和方向,从而建立时间序列有向加权复杂网络.通过对Logistic系统不同参数设置对应的时间序列复杂网络建网测试结果表明,相比经典的可视图建网方法,本文方法的网络拓扑能更简洁、直观地展示时间序列的结构特征.进而,将本文方法应用于规则排列采集的自然风场信号分析,其网络特性指标能较准确地预测采集信号的排布规律,而可视图建网方法的网络特性指标没有任何规律性的结果.     

混沌时间序列的模糊神经网络预测

设计一种新型混合模糊神经推理系统，该系统仅从期望输入输出数据集即可达到获取知识、确定模糊初始规则基的目的.再利用神经网络学习能力便不难修改规则库中的模糊规则以及隶属函数和网络权值等参数，这样大大减少了规则匹配过程，加快了推理速度，从而极大程度地提高了系统的自适应能力.用它对Mackey-Glass混沌时间序列进行预测试验,结果表明利用该网络模型无论离线还是在线学习均能对Mackey-Glass混沌时间序列进行准确的预测，证明了该系统的有效性. 关键词： 神经网络模型 模糊逻辑 混合推理系统 混沌时间序列     

Measuring time series regularity using nonlinear similarity-based sample entropy

Sampe Entropy (SampEn), a measure quantifying regularity and complexity, is believed to be an effective analyzing method of diverse settings that include both deterministic chaotic and stochastic processes, particularly operative in the analysis of physiological signals that involve relatively small amount of data. However, the similarity definition of vectors is based on Heaviside function, of which the boundary is discontinuous and hard, may cause some problems in the validity and accuracy of SampEn. Sigmoid function is a smoothed and continuous version of Heaviside function. To overcome the problems SampEn encountered, a modified SampEn (mSampEn) based on nonlinear Sigmoid function was proposed. The performance of mSampEn was tested on the independent identically distributed (i.i.d.) uniform random numbers, the MIX stochastic model, the Rossler map, and the Hennon map. The results showed that mSampEn was superior to SampEn in several aspects, including giving entropy definition in case of small parameters, better relative consistency, robust to noise, and more independence on record length when characterizing time series generated from either deterministic or stochastic system with different regularities.     

Automatic classification of killer whale vocalizations using dynamic time warping

A set of killer whale sounds from Marineland were recently classified automatically [Brown et al., J. Acoust. Soc. Am. 119, EL34-EL40 (2006)] into call types using dynamic time warping (DTW), multidimensional scaling, and kmeans clustering to give near-perfect agreement with a perceptual classification. Here the effectiveness of four DTW algorithms on a larger and much more challenging set of calls by Northern Resident whales will be examined, with each call consisting of two independently modulated pitch contours and having considerable overlap in contours for several of the perceptual call types. Classification results are given for each of the four algorithms for the low frequency contour (LFC), the high frequency contour (HFC), their derivatives, and weighted sums of the distances corresponding to LFC with HFC, LFC with its derivative, and HFC with its derivative. The best agreement with the perceptual classification was 90% attained by the Sakoe-Chiba algorithm for the low frequency contours alone.     

混沌时间序列的支持向量机预测

根据混沌动力系统的相空间延迟坐标重构理论,基于支持向量机的强大的非线性映射能力, 建立了混沌时间序列的支持向量机预测模型,并在统计学习理论的基础上采用最小二乘方法来训练预测模型,利用该模型对嵌入维数与模型的均方根误差的关系进行了探讨.最后利用Mackey-Glass时间序列和变参数的Ikeda 时间序列对该模型进行了验证,结果表明,该预测模型能精确地预测混沌时间序列,而且在混沌时间序列的嵌入维数未知时也能取得比较好的预测效果.这一结论预示着支持向量机是一种研究混沌时间序列的有效方法. 关键词： 混沌时间序列 支持向量机 最小二乘法     

Analysis of complex time series using refined composite multiscale entropy

Multiscale entropy (MSE) is an effective algorithm for measuring the complexity of a time series that has been applied in many fields successfully. However, MSE may yield an inaccurate estimation of entropy or induce undefined entropy because the coarse-graining procedure reduces the length of a time series considerably at large scales. Composite multiscale entropy (CMSE) was recently proposed to improve the accuracy of MSE, but it does not resolve undefined entropy. Here we propose a refined composite multiscale entropy (RCMSE) to improve CMSE. For short time series analyses, we demonstrate that RCMSE increases the accuracy of entropy estimation and reduces the probability of inducing undefined entropy.     

Dynamical control of systems near bifurcation points using time series

In order to excite experimentalists to apply a dynamical control method [Zhao et al., Phys. Rev. E 53, 299 (1996); 57, 5358 (1998)], we further introduce a simplified control law in this paper. The law provides a convenient way (in certain circumstance a necessary way) for experimentalists to achieve the system control when the exact position of the desired control objective cannot be known in advance. The validity of the control law is rigidly verified when the system nears a bifurcation point but our numerical examples show that it can be extended to a wide parameter region practically.     

Multiparameter estimation using only a chaotic time series and its applications

An important extension to the techniques of synchronization-based parameter estimation is presented. Based on adaptive chaos synchronization, several methods are proposed to dynamically estimate multiple parameters using only a scalar chaotic time series. In comparison with previous schemes, the presented methods decrease the cost of parameter estimation and are more applicable in practice. Numerical examples are used to demonstrate the effectiveness and robustness of the presented methods. As an example application, an implementation of multichannel digital communication is proposed, where multiparameter modulation is used to simultaneously transmit more than one digital message. From a theoretical perspective, such an encoding increases the difficulty to directly read out the message from the transmitted signal and decreases the implementation cost.     

基于高斯过程的混沌时间序列单步与多步预测

针对混沌时间序列单步和多步预测,提出基于复合协方差函数的高斯过程 (GP)模型方法.GP模型的确立由协方差函数决定,通过对训练数据集的学习,在证据最大化框架内,利用矩阵运算和优化算法自适应地确定协方差函数和均值函数中的超参数.GP模型与神经网络、模糊模型相比,其可调整参数很少.将不同复合协方差函数的GP模型应用在混沌时间序列单步及多步提前预测中,并与单一协方差函数的GP、支持向量机、最小二乘支持向量机、径向基函数神经网络等方法进行了比较.仿真结果表明,基于不同复合协方差函数的GP方法能精确地预测混沌时间序 关键词： 高斯过程 混沌时间序列 预测 模型比较     

基于混合神经网络和注意力机制的混沌时间序列预测

为提高混沌时间序列的预测精度,提出一种基于混合神经网络和注意力机制的预测模型(Att-CNNLSTM),首先对混沌时间序列进行相空间重构和数据归一化,然后利用卷积神经网络(CNN)对时间序列的重构相空间进行空间特征提取,再将CNN提取的特征和原时间序列组合,用长短期记忆网络(LSTM)根据空间特征提取时间特征,最后通过注意力机制捕获时间序列的关键时空特征,给出最终预测结果.将该模型对Logistic,Lorenz和太阳黑子混沌时间序列进行预测实验,并与未引入注意力机制的CNN-LSTM模型、单一的CNN和LSTM网络模型、以及传统的机器学习算法最小二乘支持向量机(LSSVM)的预测性能进行比较.实验结果显示本文提出的预测模型预测误差低于其他模型,预测精度更高.