Why Do Big Data and Machine Learning Entail the Fractional Dynamics?

Fractional-order calculus is about the differentiation and integration of non-integer orders. Fractional calculus (FC) is based on fractional-order thinking (FOT) and has been shown to help us to understand complex systems better, improve the processing of complex signals, enhance the control of complex systems, increase the performance of optimization, and even extend the enabling of the potential for creativity. In this article, the authors discuss the fractional dynamics, FOT and rich fractional stochastic models. First, the use of fractional dynamics in big data analytics for quantifying big data variability stemming from the generation of complex systems is justified. Second, we show why fractional dynamics is needed in machine learning and optimal randomness when asking: “is there a more optimal way to optimize?”. Third, an optimal randomness case study for a stochastic configuration network (SCN) machine-learning method with heavy-tailed distributions is discussed. Finally, views on big data and (physics-informed) machine learning with fractional dynamics for future research are presented with concluding remarks.     

自然循环流动不稳定性的多目标优化极限学习机预测方法

极限学习机是近年来提出的一种前向单隐层神经网络训练算法,具有训练速度快、不会陷入局部最优等优点,但其性能会受到随机选取的输入权值和阈值的影响.针对这一问题,提出一种基于多目标优化的改进极限学习机,将训练误差和输出层权值的均方最小化同时作为优化目标,采用带精英策略的快速非支配排序遗传算法对极限学习机的输入层到隐层的权值和阈值进行优化.将该算法应用于摇摆工况下自然循环系统不规则复合型流量脉动的多步滚动预测,分析了训练误差和输出层权值对不同步长预测效果的影响.仿真结果表明,优化极限学习机预测误差可以用较小的网络规模获得很好的泛化能力.为流动不稳定性的实时预测提供了一种准确度较高的途径,其预测结果可以作为核动力系统操作员的参考.     

Understanding the Nature of the Long-Range Memory Phenomenon in Socioeconomic Systems

In the face of the upcoming 30th anniversary of econophysics, we review our contributions and other related works on the modeling of the long-range memory phenomenon in physical, economic, and other social complex systems. Our group has shown that the long-range memory phenomenon can be reproduced using various Markov processes, such as point processes, stochastic differential equations, and agent-based models—reproduced well enough to match other statistical properties of the financial markets, such as return and trading activity distributions and first-passage time distributions. Research has lead us to question whether the observed long-range memory is a result of the actual long-range memory process or just a consequence of the non-linearity of Markov processes. As our most recent result, we discuss the long-range memory of the order flow data in the financial markets and other social systems from the perspective of the fractional Lèvy stable motion. We test widely used long-range memory estimators on discrete fractional Lèvy stable motion represented by the auto-regressive fractionally integrated moving average (ARFIMA) sample series. Our newly obtained results seem to indicate that new estimators of self-similarity and long-range memory for analyzing systems with non-Gaussian distributions have to be developed.     

A Robust Solution to Variational Importance Sampling of Minimum Variance

Importance sampling is a Monte Carlo method where samples are obtained from an alternative proposal distribution. This can be used to focus the sampling process in the relevant parts of space, thus reducing the variance. Selecting the proposal that leads to the minimum variance can be formulated as an optimization problem and solved, for instance, by the use of a variational approach. Variational inference selects, from a given family, the distribution which minimizes the divergence to the distribution of interest. The Rényi projection of order 2 leads to the importance sampling estimator of minimum variance, but its computation is very costly. In this study with discrete distributions that factorize over probabilistic graphical models, we propose and evaluate an approximate projection method onto fully factored distributions. As a result of our evaluation it becomes apparent that a proposal distribution mixing the information projection with the approximate Rényi projection of order 2 could be interesting from a practical perspective.     

How to Promote the Performance of Parametric Volatility Forecasts in the Stock Market? A Neural Networks Approach

This study uses the fourteen stock indices as the sample and then utilizes eight parametric volatility forecasting models and eight composed volatility forecasting models to explore whether the neural network approach and the settings of leverage effect and non-normal return distribution can promote the performance of volatility forecasting, and which one of the sixteen models possesses the best volatility forecasting performance. The eight parametric volatility forecasts models are composed of the generalized autoregressive conditional heteroskedasticity (GARCH) or GJR-GARCH volatility specification combining with the normal, Student’s t, skewed Student’s t, and generalized skewed Student’s t distributions. Empirical results show that, the performance for the composed volatility forecasting approach is significantly superior to that for the parametric volatility forecasting approach. Furthermore, the GJR-GARCH volatility specification has better performance than the GARCH one. In addition, the non-normal distribution does not have better forecasting performance than the normal distribution. In addition, the GJR-GARCH model combined with both the normal distribution and a neural network approach has the best performance of volatility forecasting among sixteen models. Thus, a neural network approach significantly promotes the performance of volatility forecasting. On the other hand, the setting of leverage effect can encourage the performance of volatility forecasting whereas the setting of non-normal distribution cannot.     

Skellam Type Processes of Order k and Beyond

In this article, we introduce the Skellam process of order k and its running average. We also discuss the time-changed Skellam process of order k. In particular, we discuss the space-fractional Skellam process and tempered space-fractional Skellam process via time changes in Skellam process by independent stable subordinator and tempered stable subordinator, respectively. We derive the marginal probabilities, Lévy measures, governing difference-differential equations of the introduced processes. Our results generalize the Skellam process and running average of Poisson process in several directions.     

基于Hadoop的GA-BP网络在山洪预测中的研究

研究了山洪灾害监测预警系统中雨情数据的分布式存储和分布式预测。针对采集到的水文数据急剧增长和对预测精度和预报时效的要求不断提高,分别应用Hadoop分布式文件系统对数据进行分布式存储和 MapReduce框架结合遗传算法优化神经网络的权值和阈值进行分布式预测。采用基于BP神经网络的多因子山洪灾害雨量预测模型,结合遗传算法能够实现全局优化特点来优化神经网络的权值和阈值,并在数据并行处理过程中,采用了批处理和MapReduce工作流的方式,以误差和准确率来评估预测模型,解决了神经网络在处理海量数据时训练时间长等问题。实验表明,该方法可以在不影响准确度的前提下,大大缩短运行时间,提高预测效率。     

具有增加删除机制的正则化极端学习机的混沌时间序列预测

针对正则化极端学习机的隐层具有随机选择的特性,提出了一种增加删除机制来自适应地确定正则化极端学习机的隐层节点数. 这种机制以对优化目标函数影响的大小作为评价隐层节点优劣的标准,从而淘汰那些比较“差”的节点,将那些比较“优”的节点保留下来,起到一个优化正则化极端学习机隐层节点数的目的. 与已有的只具有增加隐层节点数的机制相比较,本文提出的增加删除机制在减少正则化极端学习机隐层节点数、增强其泛化性能、提高其实时性等方面具有一定的优势. 典型混沌时间序列的实例证明了具有增加删除机制的正则化极端学习机的有效性和可行性. 关键词： 混沌时间序列 人工神经网络 极端学习机     

From Rényi Entropy Power to Information Scan of Quantum States

In this paper, we generalize the notion of Shannon’s entropy power to the Rényi-entropy setting. With this, we propose generalizations of the de Bruijn identity, isoperimetric inequality, or Stam inequality. This framework not only allows for finding new estimation inequalities, but it also provides a convenient technical framework for the derivation of a one-parameter family of Rényi-entropy-power-based quantum-mechanical uncertainty relations. To illustrate the usefulness of the Rényi entropy power obtained, we show how the information probability distribution associated with a quantum state can be reconstructed in a process that is akin to quantum-state tomography. We illustrate the inner workings of this with the so-called “cat states”, which are of fundamental interest and practical use in schemes such as quantum metrology. Salient issues, including the extension of the notion of entropy power to Tsallis entropy and ensuing implications in estimation theory, are also briefly discussed.     

The Effect of a Hidden Source on the Estimation of Connectivity Networks from Multivariate Time Series

Many methods of Granger causality, or broadly termed connectivity, have been developed to assess the causal relationships between the system variables based only on the information extracted from the time series. The power of these methods to capture the true underlying connectivity structure has been assessed using simulated dynamical systems where the ground truth is known. Here, we consider the presence of an unobserved variable that acts as a hidden source for the observed high-dimensional dynamical system and study the effect of the hidden source on the estimation of the connectivity structure. In particular, the focus is on estimating the direct causality effects in high-dimensional time series (not including the hidden source) of relatively short length. We examine the performance of a linear and a nonlinear connectivity measure using dimension reduction and compare them to a linear measure designed for latent variables. For the simulations, four systems are considered, the coupled Hénon maps system, the coupled Mackey–Glass system, the neural mass model and the vector autoregressive (VAR) process, each comprising 25 subsystems (variables for VAR) at close chain coupling structure and another subsystem (variable for VAR) driving all others acting as the hidden source. The results show that the direct causality measures estimate, in general terms, correctly the existing connectivity in the absence of the source when its driving is zero or weak, yet fail to detect the actual relationships when the driving is strong, with the nonlinear measure of dimension reduction performing best. An example from finance including and excluding the USA index in the global market indices highlights the different performance of the connectivity measures in the presence of hidden source.     

The distribution and scaling of fluctuations for Hang Seng index in Hong Kong stock market

The statistical properties of the Hang Seng index in the Hong Kong stock market are analyzed. The data include minute by minute records of the Hang Seng index from January 3, 1994 to May 28, 1997. The probability distribution functions of index returns for the time scales from 1 minute to 128 minutes are given. The results show that the nature of the stochastic process underlying the time series of the returns of Hang Seng index cannot be described by the normal distribution. It is more reasonable to model it by a truncated Lévy distribution with an exponential fall-off in its tails. The scaling of the maximium value of the probability distribution is studied. Results show that the data are consistent with scaling of a Lévy distribution. It is observed that in the tail of the distribution, the fall-off deviates from that of a Lévy stable process and is approximately exponential, especially after removing daily trading pattern from the data. The daily pattern thus affects strongly the analysis of the asymptotic behavior and scaling of fluctuation distributions. Received 9 August 2000 and Received in final form 28 August 2000     

基于拖尾分布的高分辨率合成孔径雷达图像建模

基于中心极限定理的合成孔径雷达(SAR)图像统计分布不能反映高分辨率SAR图像尖峰和厚尾的统计特征.文中使用广义中心极限定理,由雷达回波的实部和虚部的对称稳定分布,得到SAR图像的拖尾分布(幅值图像的拖尾Rayleigh分布以及强度图像的拖尾指数分布),并以拖尾Rayleigh分布为例,讨论了拖尾分布的代数拖尾特征以及尖峰厚尾的统计特性.为了实现拖尾分布对高分辨率SAR图像的精确建模,基于第二类统计量,提出了对数累积量的参数估计方法,从而高效估计出拖尾分布的参数.真实SAR图像的建模实例表明,基于广义中心极限定理的拖尾分布可以精确描述高分辨率SAR图像的尖峰和厚尾的统计特征.     

A Locally Both Leptokurtic and Fat-Tailed Distribution with Application in a Bayesian Stochastic Volatility Model

In the paper, we begin with introducing a novel scale mixture of normal distribution such that its leptokurticity and fat-tailedness are only local, with this “locality” being separately controlled by two censoring parameters. This new, locally leptokurtic and fat-tailed (LLFT) distribution makes a viable alternative for other, globally leptokurtic, fat-tailed and symmetric distributions, typically entertained in financial volatility modelling. Then, we incorporate the LLFT distribution into a basic stochastic volatility (SV) model to yield a flexible alternative for common heavy-tailed SV models. For the resulting LLFT-SV model, we develop a Bayesian statistical framework and effective MCMC methods to enable posterior sampling of the parameters and latent variables. Empirical results indicate the validity of the LLFT-SV specification for modelling both “non-standard” financial time series with repeating zero returns, as well as more “typical” data on the S&P 500 and DAX indices. For the former, the LLFT-SV model is also shown to markedly outperform a common, globally heavy-tailed, t-SV alternative in terms of density forecasting. Applications of the proposed distribution in more advanced SV models seem to be easily attainable.     

混沌映射和神经网络互扰的新型复合流密码

提出了一种将新型的神经网络互学习模型和常见的多混沌系统融合互扰的复合流密码方案. 首先利用三个Logistics混沌映射产生的随机序列作为神经网络互学习模型中三个 隐含层神经元的随机输入, 神经网络交互学习达到内部权值同步后, 再将同步权值映射为随机序列并与三个Logistics序列复合产生最终的密钥流. 实验表明, 产生的密钥流具有更好的随机性, 混沌流加密应用效果好. 关键词： 混沌映射 神经网络 权值同步 随机密钥流     

Metaheuristic optimization-based identification of fractional-order systems under stable distribution noises

This research investigates the identification problem of fractional-order chaotic systems under stable distribution noises. A powerful metaheuristic optimization method called composite differential evolution is used for the identification of the fractional-order Lorenz and Chen systems in the noisy environment, where the structure, parameters, orders and initial values of the systems are all unknown. The identification accuracy is examined when the noise follows the three special cases of stable distributions, i.e., Gaussian, Cauchy and Lévy distributions. In addition, the impact of the four parameters of stable distributions on the identification accuracy is discussed. The experimental results show that the identification error becomes larger when the noise switches from Gaussian to Cauchy and Lévy distributions. The results also turn out that the location of the stable distribution noise plays the most substantial role in the identification accuracy.     

Exact and explicit probability densities for one-sided Lévy stable distributions

We study functions gα(x) which are one-sided, heavy-tailed Lévy stable probability distributions of index α, 0<α<1, of fundamental importance in random systems, for anomalous diffusion and fractional kinetics. We furnish exact and explicit expressions for gα(x), 0 ≤ x<∞, for all α=l/k<1, with k and l positive integers. We reproduce all the known results given by k ≤ 4 and present many new exact solutions for k > 4, all expressed in terms of known functions. This will allow a "fine-tuning" of α in order to adapt gα(x) to a given experimental situation.     

Scaling of Lévy-Student processes

Student’s t-distributions are widely used in financial studies as heavy-tailed alternatives to normal distributions. As these distributions are not closed under convolution, there exist no Lévy processes with Student’s t-marginals at all points in time. In this article we show that a Student’s t-approximation of these marginals is still suitable, while not exact. Using this approximation, we are able to describe the scaling behavior of such Lévy-Student processes and the parameters of its marginal distributions by a simple analytical scaling law. This scaling law drastically simplifies the use of Lévy-Student processes as a general diffusion process in various interdisciplinary applications. We explicitly provide an application in the context of modelling high-frequency price returns.     

Stochastic Order and Generalized Weighted Mean Invariance

In this paper, we present order invariance theoretical results for weighted quasi-arithmetic means of a monotonic series of numbers. The quasi-arithmetic mean, or Kolmogorov–Nagumo mean, generalizes the classical mean and appears in many disciplines, from information theory to physics, from economics to traffic flow. Stochastic orders are defined on weights (or equivalently, discrete probability distributions). They were introduced to study risk in economics and decision theory, and recently have found utility in Monte Carlo techniques and in image processing. We show in this paper that, if two distributions of weights are ordered under first stochastic order, then for any monotonic series of numbers their weighted quasi-arithmetic means share the same order. This means for instance that arithmetic and harmonic mean for two different distributions of weights always have to be aligned if the weights are stochastically ordered, this is, either both means increase or both decrease. We explore the invariance properties when convex (concave) functions define both the quasi-arithmetic mean and the series of numbers, we show its relationship with increasing concave order and increasing convex order, and we observe the important role played by a new defined mirror property of stochastic orders. We also give some applications to entropy and cross-entropy and present an example of multiple importance sampling Monte Carlo technique that illustrates the usefulness and transversality of our approach. Invariance theorems are useful when a system is represented by a set of quasi-arithmetic means and we want to change the distribution of weights so that all means evolve in the same direction.     

基于SOM和RBF网络的软测量方法研究

针对氢粉碎过程中钕铁硼粉碎状态不可知,为有效预测合金的反应状态,提出了一种基于自组织特征映射(SOM)神经网络和径向基函数(RBF)神经网络结合构建的网络模型。在该模型中,SOM神经网络作为聚类网络,采用无教师学习算法对输入样本进行自组织分类,并将分类中心及其对应的权值向量传递给RBF神经网络,作为径向基函数的中心;RBF神经网络作为基础网络,采用高斯函数作为径向基函数实现从输入到隐含层的非线性映射,输出层则采用有教师学习算法训练网络的权值,从而实现输入层到输出层的线性映射。并以钕铁硼氢粉碎过程合金中氢含量为检测对象,运用上述方法在MATLAB平台上建立了合金中氢含量预测模型,并完成了仿真验证。     

Heavy-tailed targets and (ab)normal asymptotics in diffusive motion

We show that, under suitable confinement conditions, the ordinary Fokker-Planck equation may generate non-Gaussian heavy-tailed probability density functions (pdfs) (like, for example, Cauchy or more general Lévy stable distributions) in its long-time asymptotics. In fact, all heavy-tailed pdfs known in the literature can be obtained this way. For the underlying diffusion-type processes, our main focus is on their transient regimes and specifically the crossover features, when an initially infinite number of pdf moments decreases to a few or none at all. The time dependence of the variance (if in existence), ∼t^γ with 0<γ<2, may in principle be interpreted as a signature of subdiffusive, normal diffusive or superdiffusive behavior under confining conditions; the exponent γ is generically well defined in substantial periods of time. However, there is no indication of any universal time rate hierarchy, due to a proper choice of the driver and/or external potential.