A cycle regression analysis algorithm for extracting cycles from time-series data

A cycle regression analysis algorithm for extracting cycles from time-series data is introduced and compared against the periodogram method. Results indicate that cycle regression analysis is superior to the periodogram method. The algorithm permits the simultaneous estimation of all parameters, instead of one cycle at a time, and does not require equally spaced data. Cycle regression analysis appears to be particularly well suited to any time-series data which contain sinusoidal cycles that are related in an additive manner.     

基于子序列相似性的时间序列语义挖掘算法

时间序列是对某个事物或系统进行连续同间隔测量得到的数值序列,挖掘时间序列中潜在的语义信息对于发现系统运行规律或识别系统突发异常至关重要,然而目前多数时间序列语义挖掘算法对于时间序列数据特征有一定的约束条件,难以处理海量且特征各异的时间序列数据。针对该问题,提出一种基于子序列相似性的时间序列语义挖掘算法。通过计算子序列的相似性,将时间序列分割成片段序列进行两级聚类,识别出时间序列中潜在的物理状态。引入基于概率的迭代模式,根据候选分段情况动态调整子序列被选为参考子序列的概率,保证参考子序列涵盖全部物理状态。实验结果表明,该算法在PAMAP、Barbet等5个真实数据集上的识别准确率均超过90%,相比于FLUSS、pHMM、AutoPlait算法具有更高的识别准确率与运行效率以及更强的通用性。     

A genetic approach to the design of autonomous agents for futures trading

We propose a genetic algorithm-based method for designing an autonomous trader agent. The task of the proposed method is to find an optimal set of fuzzy if–then rules that best represents the behavior of a target trader agent. A highly profitable trader agent is used as the target in the proposed genetic algorithm. A trading history for the target agent is obtained from a series of futures trading. The antecedent part of fuzzy if–then rules considers time-series data of spot prices, while the consequent part indicates the order of trade (Buy, Sell, or No action) with its degree of certainty. The proposed method determines the antecedent part of fuzzy if–then rules. The consequent part of fuzzy if–then rules is automatically determined from the trading history of the target trader agent. The autonomous trader agent designed by the proposed genetic algorithm consists of a fixed number of fuzzy if–then rules. The decision of the autonomous trader agent is made by fuzzy inference from the time-series data of spot prices. This work was presented in part at the 11th International Symposium on Artificial Life and Robotics, Oita, Japan, January 23–25, 2006     

一种支持时间序列数据的CBR检索算法

探讨了如何为CBR（基于范例的推理）增加对一种特殊的范例类型——时间序列数据的支持．分析了基于谱分析的时间序列相似度比较算法不适用于CBR检索的缺点，并在此基础上设计了一种综合性能很好的CBR检索算法．思路是把时间序列相似度比较转化成一个卷积问题，并用DFT来简化这个卷积的计算．通过对这种CBR检索算法进行了深入的理论分析和认真的实验，结果证明，提出的算法是一个高效的算法．在这个检索算法的基础上，CBR就能够席用到时序数据的分析推理中，具有广阔的应用前景．     

Missing data simulation inside flow rate time-series using multiple-point statistics

The direct sampling (DS) multiple-point statistical technique is proposed as a non-parametric missing data simulator for hydrological flow rate time-series. The algorithm makes use of the patterns contained inside a training data set to reproduce the complexity of the missing data. The proposed setup is tested in the reconstruction of a flow rate time-series while considering several missing data scenarios, as well as a comparative test against a time-series model of type ARMAX. The results show that DS generates more realistic simulations than ARMAX, better recovering the statistical content of the missing data. The predictive power of both techniques is much increased when a correlated flow rate time-series is used, but DS can also use incomplete auxiliary time-series, with a comparable prediction power. This makes the technique a handy simulation tool for practitioners dealing with incomplete data sets.     

Detecting anomalous patterns in time-series data using sparse hierarchically parameterized transition matrices

Anomaly detection in time-series data is a relevant problem in many fields such as stochastic data analysis, quality assurance, and predictive modeling. Markov models are an effective tool for time-series data analysis. Previous approaches utilizing Markov models incorporate transition matrices (TMs) at varying dimensionalities and resolutions. Other analysis methods treat TMs as vectors for comparison using search algorithms such as the nearest neighbors comparison algorithm, or use TMs to calculate the probability of discrete subsets of time-series data. We propose an analysis method that treats the elements of a TM as random variables, parameterizing them hierarchically. This approach creates a metric for determining the “normalcy” of a TM generated from a subset of time-series data. The advantages of this novel approach are discussed in terms of computational efficiency, accuracy of anomaly detection, and robustness when analyzing sparse data. Unlike previous approaches, this algorithm is developed with the expectation of sparse TMs. Accounting for this sparseness significantly improves the detection accuracy of the proposed method. Detection rates in a variety of time-series data types range from (97 % TPR, 2.1 % FPR) to (100 % TPR, <0.1 % FPR) with very small sample sizes (20–40 samples) in data with sparse transition probability matrices.     

基于卷积的时态CBR快速检索算法及应用*

探讨了如何增强CBR对一种常见的时态信息,即时间序列数据的检索能力;分析了已有的基于傅里叶频谱分析的时间序列检索算法应用于CBR时遇到的问题,并根据时态CBR检索的需要,提出了一种新的基于循环卷积和傅里叶变换时间序列检索算法.理论分析和数值实验结果都证明,提出的算法在检索效率上有一定的优势.将采取这种检索方法的时态CBR应用于时间序列的预测问题中,取得了较好的预测效果且具有较高的预测效率.     

交互更新模式的量子遗传算法的几何约束求解

目的针对传统量子遗传算法无法充分利用种群中未成熟个体信息的不足,提出了基于交互更新模式的量子遗传算法(IUMQGA)并应用于几何约束求解中。方法几何约束问题的约束方程组可转化为优化模型,因此约束求解问题可以转化为优化问题。采用将遗传算法与量子理论相结合的量子遗传算法,使用双串量子染色体结构,使用交互更新策略将遗传算法中的交叉操作利用量子门变换来实现,根据不同情况采用不同的交互更新策略。这里的交互,指的是两个个体进行信息交换的过程,该过程用以产生新的个体。这不仅增加了个体间信息的交换而且充分利用了种群中未成熟个体的信息,提高了算法的收敛速度。结果通过非线性方程实例和几何约束实例测试并与其他方法比较表明,基于交互更新模式的量子遗传算法求解几何约束问题具有更好的求解精度和求解速率。双圆外公切线问题实例中,IUMQGA算法比QGA算法稳定;单圆填充问题和双圆外公切线问题实例中,通过实验求得各变量的最优值与其相应的精确值的误差在1E-2以下。结论采用交互更新模式的量子遗传算法可以很好地求解几何约束问题。     

时序规则挖掘

提出了新颖的时间序列模式和规则挖掘技术。该技术先把待挖掘的时间序列转换成子时间序列数据,然后利用子时间序列所隐藏的知识,来指导对原时间序列的挖掘,从中提取模式或规则。给出了时间序列模式和规则的挖掘算法,并举例说明该算法是有效和可行的。     

Distortion-free predictive streaming time-series matching

Efficient processing of streaming time-series generated by remote sensors and mobile devices has become an important research area. As in traditional time-series applications, similarity matching on streaming time-series is also an essential research issue. To obtain more accurate similarity search results in many time-series applications, preprocessing is performed on the time-series before they are compared. The preprocessing removes distortions such as offset translation, amplitude scaling, linear trends, and noise inherent in time-series. In this paper, we propose an algorithm for distortion-free predictive streaming time-series matching. Similarity matching on streaming time-series is saliently different from traditional time-series in that it is not feasible to directly apply the traditional algorithms for streaming time-series. Our algorithm is distortion-free in the sense that it performs preprocessing on streaming time-series to remove offset translation and amplitude scaling distortions at the same time. Our algorithm is also predictive, since it performs streaming time-series matching against the predicted most recent subsequences in the near future, and thus improves search performance. To the best of our knowledge, no streaming time-series matching algorithm currently performs preprocessing and predicts future search results simultaneously.     

基于自适应遗传算法的股票预测模型研究

为了解决单一神经网络模型很难满足股票预测建模要求的问题,提出一种基于遗传算法的粗糙集属性约简方法和神经网络相结合的预测模型。在该模型中,改进了自适应性遗传算法的交叉算子与变异算子。基于该遗传算法的粗糙集属性约简相比传统的粗糙集属性约简,其具有更强的求解最小属性约简的能力,解决了神经网络预测时训练速度慢、内存开销大等问题;在数据预处理过程中,引入聚类分析,有效解决了连续属性离散化的问题。实验结果证明,该预测模型具有较高的预测精度,在时间序列的股票预测中是相当有效的。     

A linear-time model-checking algorithm for the alternation-free modal mu-calculus

We develop a model-checking algorithm for a logic that permits propositions to be defined using greatest and least fixed points of mutually recursive systems of equations. This logic is as expressive as the alternation-free fragment of the modal mu-calculus identified by Emerson and Lei, and it may therefore be used to encode a number of temporal logics and behavioral preorders. Our algorithm determines whether a process satisfies a formula in time proportional to the product of the sizes of the process and the formula; this improves on the best known algorithm for similar fixed-point logics.     

基于Mult-TWDTW算法的时序SAR图像土地覆盖分类

合成孔径雷达(SAR)图像为土地覆盖分类提供了重要的时序数据源. 现有的时间序列匹配算法可以充分挖掘时序特征的相似性信息, 从而获得较好的分类效果. 本文引入了综合考虑形状相似性和物候差异的经典时序匹配算法TWDTW (time weighted dynamic time warping)指导SAR土地覆盖分类, 并针对传统TWDTW仅考虑单一特征时间序列上的相似性匹配问题, 提出了一种基于多特征联合的时间加权动态时间规整算法(Mult-TWDTW). 该方法首先提取后向散射系数、干涉相干性以及双极化雷达植被指数(dual polarization radar vegetation Index, DpRVI) 这3种特征, 然后在TWDTW算法基础上联合多个特征设计了Mult-TWDTW模型. 为验证所提方法的有效性, 使用Sentinel-1A时序数据在丹江口区域完成土地覆盖分类, 并将Mult-TWDTW与MLP、1D-CNN、K-means、SVM和使用单特征的TWDTW算法进行对比. 实验结果显示, Mult-TWDTW算法得到了最好的分类效果, 总体精度和Kappa系数可以达到95.09%和91.76, 表明Mult-TWDTW算法有效联合了多个特征信息, 能够提升时序匹配算法在多种土地覆盖类别分类中的潜力.     

基于多边形边界约简的启发式在线时间序列表示算法

提出了一种新颖的可在线计算的时间序列启发式算法。算法具有多边形约简算法相同的优良的近似质量，并可在固定数据缓冲区空间内在线运算。用启发式搜索方法自动获取最佳分段数。在随机时间序列上仿真试验证明算法有很高的逼近质量和较低的计算复杂性。     

Application of Computational Mechanics to the Analysis of Seismic Time-series via Numerical Optimisation

We apply the Computational Mechanics approach to the analysis of time-series representative of geophysical measurements. The algorithm employed is the Causal-State Splitting Reconstruction (CSSR) algorithm. We address a number of data pre-processing steps which are necessary when analysing complex time-series and specific to symbolised time-series analysis tools such as CSSR. We cast the choice of input parameters for the CSSR algorithm and time-series symbolisation, into an optimisation problem, with the aim of maximising the predictability of events of specific interest. Our approach is problem independent and can be easily extended to other applications. This research highlights the challenges to be overcome when analysing complex time-series using the Computational Mechanics approach. We also discuss further developments necessary to extend the approach to real data applications.

Inductive process modeling

In this paper, we pose a novel research problem for machine learning that involves constructing a process model from continuous data. We claim that casting learned knowledge in terms of processes with associated equations is desirable for scientific and engineering domains, where such notations are commonly used. We also argue that existing induction methods are not well suited to this task, although some techniques hold partial solutions. In response, we describe an approach to learning process models from time-series data and illustrate its behavior in three domains. In closing, we describe open issues in process model induction and encourage other researchers to tackle this important problem. Editor: David Page.     

Unsupervised feature selection for sensor time-series in pervasive computing applications

The paper introduces an efficient feature selection approach for multivariate time-series of heterogeneous sensor data within a pervasive computing scenario. An iterative filtering procedure is devised to reduce information redundancy measured in terms of time-series cross-correlation. The algorithm is capable of identifying nonredundant sensor sources in an unsupervised fashion even in presence of a large proportion of noisy features. In particular, the proposed feature selection process does not require expert intervention to determine the number of selected features, which is a key advancement with respect to time-series filters in the literature. The characteristic of the prosed algorithm allows enriching learning systems, in pervasive computing applications, with a fully automatized feature selection mechanism which can be triggered and performed at run time during system operation. A comparative experimental analysis on real-world data from three pervasive computing applications is provided, showing that the algorithm addresses major limitations of unsupervised filters in the literature when dealing with sensor time-series. Specifically, it is presented an assessment both in terms of reduction of time-series redundancy and in terms of preservation of informative features with respect to associated supervised learning tasks.     

A UML profile for the conceptual modelling of data-mining with time-series in data warehouses

Time-series analysis is a powerful technique to discover patterns and trends in temporal data. However, the lack of a conceptual model for this data-mining technique forces analysts to deal with unstructured data. These data are represented at a low-level of abstraction and their management is expensive. Most analysts face up to two main problems: (i) the cleansing of the huge amount of potentially-analysable data and (ii) the correct definition of the data-mining algorithms to be employed. Owing to the fact that analysts’ interests are also hidden in this scenario, it is not only difficult to prepare data, but also to discover which data is the most promising. Since their appearance, data warehouses have, therefore, proved to be a powerful repository of historical data for data-mining purposes. Moreover, their foundational modelling paradigm, such as, multidimensional modelling, is very similar to the problem domain. In this article, we propose a unified modelling language (UML) extension through UML profiles for data-mining. Specifically, the UML profile presented allows us to specify time-series analysis on top of the multidimensional models of data warehouses. Our extension provides analysts with an intuitive notation for time-series analysis which is independent of any specific data-mining tool or algorithm. In order to show its feasibility and ease of use, we apply it to the analysis of fish-captures in Alicante. We believe that a coherent conceptual modelling framework for data-mining assures a better and easier knowledge-discovery process on top of data warehouses.     

一种快速的子序列匹配算法

提出一种在时间序列上快速匹配子序列的算法,该算法不同于FRM算法,而是采用VA-file这种索引结构,将数据点直接存储在索引上,并在该索引的基础上设计了一种进行范围查询的方法.实验采用了三种时间序列数据集,从不同的角度验证算法的有效性,结果表明该算法大大提高了查询性能.     

面向多元时间序列的群体因果关系发现算法

从多元时间序列观测数据中学习多个变量之间的因果关系是许多专业领域中的重要基本问题。现有的多元时间序列因果关系发现方法通常从每个个体的观测数据中学习个体因果关系,没有考虑部分个体之间可能存在相同的因果关系,导致样本利用不足。提出一种面向多元时间序列的群体因果关系发现算法。该算法分为2个阶段：第一阶段基于因果关系对个体之间的相似性进行度量,并把多个个体划分成多个群体,且无须指定群体的个数;第二阶段基于变分推断方法充分利用每个群体内的所有个体数据,从而学习群体因果关系。实验结果表明,该算法在多组不同参数生成的仿真数据上均具有较好的表现,与对比算法相比,AUC评分提升了5%～20%。在真实数据集中,该算法能够较好地区分具有不同因果关系的群体,并且能够学习到不同群体之间不同的因果关系,表明算法不仅具有因果关系发现能力,而且还具有多元时间序列聚类能力。