Time series classification by class-specific Mahalanobis distance measures

To classify time series by nearest neighbors, we need to specify or learn one or several distance measures. We consider variations of the Mahalanobis distance measures which rely on the inverse covariance matrix of the data. Unfortunately??for time series data??the covariance matrix has often low rank. To alleviate this problem we can either use a pseudoinverse, covariance shrinking or limit the matrix to its diagonal. We review these alternatives and benchmark them against competitive methods such as the related Large Margin Nearest Neighbor Classification (LMNN) and the Dynamic Time Warping (DTW) distance. As we expected, we find that the DTW is superior, but the Mahalanobis distance measures are one to two orders of magnitude faster. To get best results with Mahalanobis distance measures, we recommend learning one distance measure per class using either covariance shrinking or the diagonal approach.     

Exploiting statistical properties of wavelet coefficient for face detection and recognition

Wavelet transforms (WT) are widely accepted as an essential tool for image processing and analysis. Image and video compression, image watermarking, content-base image retrieval, face recognition, texture analysis, and image feature extraction are all but few examples. It provides an alternative tool for short time analysis of quasi-stationary signals, such as speech and image signals, in contrast to the traditional short-time Fourier transform. The Discrete Wavelet Transform (DWT) is a special case of the WT, which provides a compact representation of a signal in the time and frequency domain. In particular, wavelet transforms are capable of representing smooth patterns as well anomalies (e.g. edges and sharp corners) in images. We are focusing here on using wavelet transforms statistical properties for facial feature detection, which allows us to extract the image facial feature/edges easily. Wavelet sub-bands segmentation method been developed and used to clean up the non-significant wavelet coefficients in wavelet sub-band (k) based on the (k-1) sub-band. Moreover, erosion which is considered as one of the fundamental operation in morphological image processing, been used to reduce the unwanted edges in certain directions. For face detection, face template profiles been built for both the face and the eyes for different wavelet sub-band levels to achieve better computational performance, these profiles used to match the extracted profiles from the wavelet domain of the input image using the Dynamic Time Warping technique DTW. The DTW smallest distance allows identifying the face and the eyes location. The performance of face features distances and ratio has been also tested for face verification purposes. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

A multi-ary number communication system based on hyperchaotic system of 6th-order cellular neural network

This paper proposes the hyperchaotic system of 6th-order cellular neural network (CNN), realizes its synchronization based on state observer. In addition, a multi-ary number communication system based on this hyperchaotic system is given in this paper. This communication system has the features of large capacity of signals transmission and high security.     

Hardware Implementation of Discrete Stochastic Arithmetic

In this paper we present a hardware implementation of the Discrete Stochastic Arithmetic (DSA) which is based on CESTAC (Controle et Estimation STochastique des Arrondis de Calculs), a method of controlling round-off errors in floating-point scientific computations. Real-time software implementation of this method suffers from computation bottlenecks. This paper gives a hardware alternative that would significantly accelerate the computation. The proposed architecture is based on a Stochastic Floating-Point Unit (SFPU) which performs discrete stochastic operations. This SFPU has been integrated in a coprocessor, used in a complete System on Chip (SoC).     

Divide-and-conquer memetic algorithm for online multi-objective test paper generation

Online test paper generation (Online-TPG) generates a test paper automatically online according to user specification based on multiple assessment criteria, and the generated test paper can then be attempted online by user. Online-TPG is challenging as it is a multi-objective optimization problem that is NP-hard, and it is also required to satisfy the online generation requirement. In this paper, we propose an efficient multi-objective optimization approach based on the divide-and-conquer memetic algorithm (DAC-MA) for Online-TPG. Instead of solving the multi-objective constraints simultaneously, the set of constraints is divided into two subsets of relevant constraints, which can then be solved separately and effectively by evolutionary computation and local search of DAC-MA. The empirical performance results have shown that the proposed approach has outperformed other TPG techniques in terms of runtime efficiency and paper quality.     

云计算环境下人工蜂群作业调度算法设计

针对云计算环境下作业调度优化问题,提出了一种基于人工蜂群的调度算法.分析人工蜂群算法的求解组合优化问题过程,建立了收益度函数和蜜源位置更新公式,最后论述了利用该算法求解的具体步骤.并通过实验分析了该算法的性能.     

Cluster based branching for the asymmetric traveling salesman problem

This paper presents a new branching scheme for the asymmetric traveling salesman problem (ATSP) based on clusters. A cluster is defined as a node set with the characteristic that there exists an optimal solution in which the nodes in the node set are visited consecutively. The paper considers identification of clusters, implementation of a cluster based branching scheme, and cluster based dominance tests. The new approach is implemented in a branch and bound algorithm using a well-known additive bounding procedure. Considerable savings in computing time are obtained compared to previously published assignment based branch and bound algorithms for the ATSP.     

Advanced Genetic Programming Based Machine Learning

A Genetic Programming based approach for solving classification problems is presented in this paper. Classification is understood as the act of placing an object into a set of categories, based on the object’s properties; classification algorithms are designed to learn a function which maps a vector of object features into one of several classes. This is done by analyzing a set of input-output examples (“training samples”) of the function. Here we present a method based on the theory of Genetic Algorithms and Genetic Programming that interprets classification problems as optimization problems: Each presented instance of the classification problem is interpreted as an instance of an optimization problem, and a solution is found by a heuristic optimization algorithm. The major new aspects presented in this paper are advanced algorithmic concepts as well as suitable genetic operators for this problem class (mainly the creation of new hypotheses by merging already existing ones and their detailed evaluation). The experimental part of the paper documents the results produced using new hybrid variants of Genetic Algorithms as well as investigated parameter settings. Graphical analysis is done using a novel multiclass classifier analysis concept based on the theory of Receiver Operating Characteristic curves. The work described in this paper was done within the Translational Research Project L282 “GP-Based Techniques for the Design of Virtual Sensors” sponsored by the Austrian Science Fund (FWF).     

Feasible smooth method based on Barzilai–Borwein method for stochastic linear complementarity problem

In this paper, we propose a feasible smooth method based on Barzilai–Borwein (BB) for stochastic linear complementarity problem. It is based on the expected residual minimization (ERM) formulation for the stochastic linear complementarity problem. Numerical experiments show that the method is efficient.     

Noisy image segmentation based on nonlinear diffusion equation model

This paper addresses the segmentation problem in noisy image based on nonlinear diffusion equation model and proposes a new adaptive segmentation model based on gray-level image segmentation model. This model also can be extended to the vector value image segmentation. By virtue of the prior information of regions and boundary of image, a framework is established to construct different segmentation models using different probability density functions. A segmentation model exploiting Gauss probability density function is given in this paper. An efficient and unconditional stable algorithm based on locally one-dimensional (LOD) scheme is developed and it is used to segment the gray image and the vector values image. Comparing with existing classical models, the proposed approach gives the best performance.     

基于FI-代数的一个逻辑系统

建立一种基于FI-代数的模糊命题演算的形式演绎系统．并讨论了该系统语义的完备性。其目的在于使通常的众多模糊推理系统能够纳入该逻辑系统之中．以便在一个更加广泛的代数和逻辑框架下来研究模糊推理的逻辑基础问题。     

Formalization of implication based fuzzy reasoning method

Fuzzy reasoning includes a number of important inference methods for addressing uncertainty. This line of fuzzy reasoning forms a common logical foundation in various fields, such as fuzzy logic control and artificial intelligence. The full implication triple I method (a method only based on implication, TI method for short) for fuzzy reasoning is proposed in 1999 to improve the popular CRI method (a hybrid method based on implication and composition). The current paper delves further into the TI method, and a sound logical foundation is set for the TI method based on the monoidal t-norm based logical system MTL.     

基于缓冲超越概率的风险度量及套期保值

尾部风险测度是风险管理中的关键点,本文利用缓冲超越概率模型,量化不同预期损失的风险概率分布,构建条件风险价值约束下的最小化“厚尾事件”概率的套期保值策略,从而将现有研究的视角拓展到考虑预期损失和风险概率的双重维度。本文通过实证数据统计和参数化拟合分布两个方法提供不同风险阈值及对应的缓冲超越概率的稳定解集合,研究结果发现,无论预期损失服从厚尾分布还是正态分布,缓冲超越概率模型均能够显著地降低市场风险和潜在的“厚尾事件”发生的概率,并提供比最小化方差稳定的套期保值比率。     

Interval estimation of the mean of a normal distribution based on quantized observations

We consider the problem of making statistical inference about the mean of a normal distribution based on a random sample of quantized (digitized) observations. This problem arises, for example, in a measurement process with errors drawn from a normal distribution and with a measurement device or process with a known resolution, such as the resolution of an analog-to-digital converter or another digital instrument. In this paper we investigate the effect of quantization on subsequent statistical inference about the true mean. If the standard deviation of the measurement error is large with respect to the resolution of the indicating measurement device, the effect of quantization (digitization) diminishes and standard statistical inference is still valid. Hence, in this paper we consider situations where the standard deviation of the measurement error is relatively small. By Monte Carlo simulations we compare small sample properties of the interval estimators of the mean based on standard approach (i.e. by ignoring the fact that the measurements have been quantized) with some recently suggested methods, including the interval estimators based on maximum likelihood approach and the fiducial approach. The paper extends the original study by Hannig et al. (2007).     

SETAR model selection-A bootstrap approach

Summary  The aim of this paper is to propose new selection criteria for the orders of selfexciting threshold autoregressive (SETAR) models. These criteria use bootstrap methodology; they are based on a weighted mean of the apparent error rate in the sample and the average error rate obtained from bootstrap samples not containing the point being predicted. These new criteria are compared with the traditional ones based on the Akaike information criterion (AIC). A simulation study and an example on a real data set end the paper.     

On a student-related optimal control problem

In a recent paper (Ref. 1), Cheng and Teo discussed some further extensions of a student-related optimal control problem which was originally proposed by Raggettet al. (Ref. 2) and later on modified by Parlar (Ref. 3). In this paper, we treat further extensions of the problem.This paper is a modified and improved version of Ref. 4. It is based, in part, on research sponsored by NSF.     

交通流模型基于特征投影分解技术的外推降维有限差分格式

利用特征正交分解(proper orthogonal decomposition,简记为POD)技术研究交通流的Aw-Rascle-Zhang(ARZ)模型. 建立一种基于 POD方法维数较低的外推降维有限差分格式, 并用数值例子检验数值计算结果与理论结果相吻合, 进一步表明基于POD方法的外推降维有限差分格式对于求解交通流方程数值解是可行和有效的.     

移动传感器网络中的最大价值路径扫描覆盖算法

扫描覆盖是当前移动传感器网络的一个重要覆盖技术,其主要通过规划移动传感器的巡逻路径对事件兴趣点（Points of Interest,POI）进行定期监测,从而以相对于普通覆盖方案更低廉的成本实现对POI监控．研究最大价值路径扫描覆盖,即使用移动传感器扫描覆盖分布在一条路径上的POI集合,使得被覆盖POI的价值总和达到最大．首先设计了一个基于线性规划随机取整的近似算法,通过将问题松弛并刻画为一个线性规划,然后对线性规划最优解取整得到一个扫描覆盖方案．该算法可在O（mn^3.5L）时间内求解,并具有可证明的近似比1-1/e．其次,通过扩展基于贪心策略的集合覆盖算法,设计了一个时间复杂度为O（m²n²）的贪心算法,其主要思想为循环选取一个单位巡逻范围覆盖POI价值最大的传感器．为优化运行时间,基于MVSCP问题的特殊结构将算法时间进一步改进至O（m log m+mn²）．最后,通过仿真实验分析所设计算法的实际性能.实验结果表明,线性规划随机取整算法运行时间低至整数规划算法的百分之一,但其所求解的质量只略低于整数规划算法;改进的贪心算法虽然不具有可证明的近似比,但其实际所求解的质量并不弱于线性规划随机取整算法,并且具有三者中最佳的运行时间．     

SOS: A Quantile Estimation Procedure for Dynamic Lot-sizing Problems

This paper reviews some of the current approaches available for computing the demand quantiles required to plan the procurement of items with stochastic non-stationary demands. The paper first describes the stochastic single-item lot-sizing problem considered and then presents a practical solution approach based on a dynamic lot-sizing model. Three methods available to compute demand quantiles are then reviewed and a new procedure based on smoothed order statistics (SOS) is proposed. Finally, the behaviour of these estimation methods, when used to solve single-item lot-sizing problems with non-stationary stochastic demands, is studied by simulation.     

微观交通仿真与GIS数据共享的本体方法

针对微观交通仿真与GIS数据共享存在的数据冗余和低效的问题,基于本体理论,提出微观交通仿真与GIS数据共享框架,给出数据共享框架的分层设计.对微观交通仿真数据特征进行概括和抽象,研究并设计微观交通仿真本体.提出基于迭代思想的微观交通仿真本体(MTSON)构建方法,重点阐述微观交通仿真本体的需求获取、分析、设计与实现过程.最后以OWL为形式化描述语言,对获取的微观交通仿真本体进行表达,为微观交通仿真与GIS数据共享的平滑实现奠定了基础.