Detecting change-points in Markov chains

Markov chains provide a flexible model for dependent random variables with applications in such disciplines as physics, environmental science and economics. In the applied study of Markov chains, it may be of interest to assess whether the transition probability matrix changes during an observed realization of the process. If such changes occur, it would be of interest to estimate the transitions where the changes take place and the probability transition matrix before and after each change. For the case when the number of changes is known, standard likelihood theory is developed to address this problem. The bootstrap is used to aid in the computation of p-values. When the number of changes is unknown, the AIC and BIC measures are used for model selection. The proposed methods are studied empirically and are applied to example sets of data.     

A reactive GRASP and path relinking for a combined production–distribution problem

An NP-hard production–distribution problem for one product over a multi-period horizon is investigated. The aim is to minimize total cost taking production setups, inventory levels and distribution into account. An integer linear model is proposed as a compact problem specification but it cannot be solved to optimality for large instances. Instead of using a classical two-phase approach (production planning and then route construction for each day), metaheuristics that simultaneously tackle production and routing decisions are developed: a GRASP (greedy randomized adaptive search procedure) and two improved versions using either a reactive mechanism or a path-relinking process. These algorithms are evaluated on 90 randomly generated instances with 50, 100 and 200 customers and 20 periods. The results confirm the interest of integrating production and distribution decisions, compared to classical two-phase methods. Moreover, reaction and path-relinking give better results than the GRASP alone.     

中式卷烟包装设计之管窥

本文以《烟草控制框架公约》在我国正式生效后,按照《公约》对卷烟包装的规定,我国的卷烟商标和包装在很大程度上要重新设计和制作为着眼点,从正确履行《烟草控制框架公约》保持中式卷烟包装设计特色、正确理解中式卷烟内涵,明确烟标和包装设计方向、中式卷烟包装和色彩与消费心理有着密切的关系以及中式卷烟包装应该提高环保意识、减少环境污染和中式卷烟包装应提高技术含量,利用核心技术进行防伪设计等方面,论述了中式卷烟包装设计的发展方向及注意的事项。     

Privacy-preserving boosting

We describe two algorithms, BiBoost (Bipartite Boosting) and MultBoost (Multiparty Boosting), that allow two or more participants to construct a boosting classifier without explicitly sharing their data sets. We analyze both the computational and the security aspects of the algorithms. The algorithms inherit the excellent generalization performance of AdaBoost. Experiments indicate that the algorithms are better than AdaBoost executed separately by the participants, and that, independently of the number of participants, they perform close to AdaBoost executed using the entire data set. Responsible Editor: Charu Aggarwal.     

Limitations of nonlinear discrete periodic control for disturbance attenuation and robust stabilization

This paper presents a performance analysis of nonlinear periodically time-varying discrete controllers acting upon a linear time-invariant discrete plant. Time-invariant controllers are distinguished from strictly periodically time-varying controllers. For a given nonlinear periodic controller, a time-invariant controller is constructed. Necessary and sufficient conditions are given under which the time-invariant controller gives strictly better control performance than the time-invariant controller from which it was obtained, for the attenuation of l_p exogenous disturbances and the robust stabilization of l_p unstructured perturbations, for all p[1,∞].     

On the bounded-real and positive-real conditions for affine nonlinear state-delayed systems and applications to stability

In this paper, bounded-real conditions for affine nonlinear state-delayed systems are derived using the concept of dissipativeness. Necessary and sufficient conditions for the system to be dissipative and to have finite L₂-gain also referred to as the bounded-real condition are given. The implications on the stability of the system and feedback interconnections of such systems are also considered. Finally, an equivalent of the positive-real lemma is derived and its implications on stability of the system and feedback interconnections of such systems are similarly discussed.     

A method of predicting the number of clusters using Rand's statistic

Distributional and asymptotic results on the moment of Rand's C_k statistic were derived by DuBien and Warde [1981. Some distributional results concerning a comparative statistic used in cluster analysis. ASA Proceedings of the Social Statistics Section, 309–313.]. Based on those results, a method to predict the number of clusters is suggested by applying various agglomerative clustering algorithms. In the procedure, the methods using different indexes are examined and compared based on the concept of agreement (or, disagreement) between clusterings generated by different clustering algorithms on the set of data. Our method having practical generality works better than the other methods and assigns statistical meaning to C_k values in determining the number of clusters from the comparison.     

Fuzzy medial axis transformation (fmat): Image representation,skeleton extraction and uncertainty management in computer vision

The problem of extraction of medial axis transformation of a gray image with reference to skeletonization, image representation and uncertainty management in a vision system has been addressed. The Fuzzy Medial Axis Transformation (fmat) of a fuzzy setf is a set of fuzzy disks whose sup isf. Unfortunately, specifying thefmat sometimes requires more storage space than specifyingf itself. The present paper describes some techniques to improve the compact representation offmat; thereby making it practically useful to an image for its skeleton extraction and compact representation, for shape analysis and template matching, for representation and retrieval, for uncertainty management in recognition and for creating new images of various poses. The algorithms involve reduction of redundancy infmat, its approximation, and reduction of the searching spaces for its computation. Computational aspects for the convenience of writing an efficient program have been described. Some applications of thefmat have also been mentioned. This work was done while SKP held annrc-nasa Senior Research Associateship at the Lyndon B Johnson Space Center, Houston, Texas.     

Elastic properties of As-Sb-Se glasses

Results of measurement of elastic modulii on As-Sb-Se glasses are reported and their composition dependence discussed. The Young’s and the shear modulii lie in the range of 170–210 and 65–80 kb respectively. These values are typical of chalcogenide glasses. For (As, Sb)₄₀Se₆₀ glasses, the modulii increase monotonically with increasing Sb₂Se₃ content. The observed composition dependence of the modulii for the As _x Sb₁₅Se_{85 −x} glasses is examined in terms of the chemically ordered structural units in the glasses.     

Chemical amplification — A novel approach to ultratrace analysis

The lowest concentration that can be estimated with 100% error in chemical analysis by instrumental technique is referred to as detection limit. It is equal to the ratio of twice the background signal to the magnitude of analytical signal corresponding to a concentration of unity. Detection limit (dl) could be improved if the analytical signal alone could be selectively amplified without affecting the background. It is shown that this could be achieved by chemical amplification. Atomic amplification of two or three orders of magnitude could be obtained by combining two chemical amplifications in series. It is demonstrated that by using heteropoly acid formation as the first amplification reaction and determining molybdenum in the heteropoly acid by catalyticdc polarographic wave, a detection limit for the hetero atom of less than a ppb can be obtained. It is pointed out that the approach of selectively amplifying analytical signals by tandem chemical amplification reactions makes it possible to perform ultratrace analysis (ppb level) with classical instrumental analytical techniques which havedl at ppm level.