Geomechanical parameters identification by particle swarm optimization and support vector machine

Back analysis is commonly used in identifying geomechanical parameters based on the monitored displacements. Conventional back analysis method is not capable of recognizing non-linear relationship involving displacements and mechanical parameters effectively. The new intelligent displacement back analysis method proposed in this paper is the combination of support vector machine, particle swarm optimization, and numerical analysis techniques. The non-linear relationship is efficiently represented by support vector machine. Numerical analysis is used to create training and testing samples for recognition of SVMs. Then, a global optimum search on the obtained SVMs by particle swarm optimization can lead to the geomechanical parameters identification effectively.     

Ontological approach to development of computing with words based systems

Computing with words introduced by Zadeh becomes a very important concept in processing of knowledge represented in the form of propositions. Two aspects of this concept – approximation and personalization – are essential to the process of building intelligent systems for human-centric computing.For the last several years, Artificial Intelligence community has used ontology as a means for representing knowledge. Recently, the development of a new Internet paradigm – the Semantic Web – has led to introduction of another form of ontology. It allows for defining concepts, identifying relationships among these concepts, and representing concrete information. In other words, an ontology has become a very powerful way of representing not only information but also its semantics.The paper proposes an application of ontology, in the sense of the Semantic Web, for development of computing with words based systems capable of performing operations on propositions including their semantics. The ontology-based approach is very flexible and provides a rich environment for expressing different types of information including perceptions. It also provides a simple way of personalization of propositions. An architecture of computing with words based system is proposed. A prototype of such a system is described.     

隐马尔可夫模型在语音识别中的应用

隐马尔可夫模型 (HMM)的基本技术是语音识别中较为成功的算法 .主要是它具有较强的对时间序列结构的建模能力 .本文首先深入浅出地介绍了 HMM的基本技术和一个基于 HMM的孤立词语音识别系统的构成方法 ,其次 ,基于 HMM尚存有一些缺陷 ,造成语音识别能力较弱 ,为此本文又进一步阐述了语音识别应用中的几种改进的 HMM系统及目前的热点方法—— HMM与 ANN构成的混合网络     

GA-based learning for a model-based object recognition system

This paper proposes a genetic-algorithm-based learning strategy that models membership functions of the fuzzy attributes of surfaces in a model based machine vision system. The objective function aims at enhancing recognition performance in terms of maximizing the degree of discrimination among classes. As a result, the accuracy of recognizing known instances of objects and generalization capability by recognizing unknown instances of known objects are greatly improved. Performance enhancement is achieved by incorporating an off-line learning mechanism using genetic algorithm in the feedback path of the recognition system.     

Minimum classification error training in example based speech and pattern recognition using sparse weight matrices

The Minimum Classification Error (MCE) criterion is a well-known criterion in pattern classification systems. The aim of MCE training is to minimize the resulting classification error when trying to classify a new data set. Usually, these classification systems use some form of statistical model to describe the data. These systems usually do not work very well when this underlying model is incorrect. Speech recognition systems traditionally use Hidden Markov Models (HMM) with Gaussian (or Gaussian mixture) probability density functions as their basic model. It is well known that these models make some assumptions that are not correct. In example based approaches, these statistical models are absent and are replaced by the pure data. The absence of statistical models has created the need for parameters to model the data space accurately. For this work, we use the MCE criterion to create a system that is able to work together with this example based approach. Moreover, we extend the locally scaled distance measure with sparse, block diagonal weight matrices resulting in a better model for the data space and avoiding the computational load caused by using full matrices. We illustrate the approach with some example experiments on databases from pattern recognition and with speech recognition.     

Dynamic construction of statistical models in managerial DSS

The fields of operations research (OR) and artificial intelligence (AI) provide complementary methods that may be combined into managerial decision support systems (DSS). However, the management domain is substantially different from domains in which prior expert systems have been developed. Consequently, successful application of OR/AI techniques in managerial DSS requires careful analysis and additional development. Ongoing research concerning design and implementation of managerial DSS is discussed. A prototype system capable of constructing linear statistical models of direct and indirect relationships from a knowledge base of relationships is described and evaluated.     

Effect of random parameter switching on commensurate fractional order chaotic systems

The paper explores the effect of random parameter switching in a fractional order (FO) unified chaotic system which captures the dynamics of three popular sub-classes of chaotic systems i.e. Lorenz, Lu and Chen's family of attractors. The disappearance of chaos in such systems which rapidly switch from one family to the other has been investigated here for the commensurate FO scenario. Our simulation study show that a noise-like random variation in the key parameter of the unified chaotic system along with a gradual decrease in the commensurate FO is capable of suppressing the chaotic fluctuations much earlier than that with the fixed parameter one. The chaotic time series produced by such random parameter switching in nonlinear dynamical systems have been characterized using the largest Lyapunov exponent (LLE) and Shannon entropy. The effect of choosing different simulation techniques for random parameter FO switched chaotic systems have also been explored through two frequency domain and three time domain methods. Such a noise-like random switching mechanism could be useful for stabilization and control of chaotic oscillation in many real-world applications.     

On solvability recognition for interval linear systems of equations

The paper considers the problem of recognizing solvability (nonemptiness of the solution set) for interval systems of linear algebraic equations. We introduce a quantitative measure of the membership of a point in the solution set, the so-called “recognizing functional” of the solution set. As the result, the decision on solvability of the interval linear systems reduces to global maximization of the recognizing functional. Additionally, the specific value of this maximum and its argument provide us with important quantitative information of the solvability supply or its deficiency, which can used for the correction of the interval system in a desired sense.     

Linearly Implicit Finite Element Methods for the Time-Dependent Joule Heating Problem

Completely discrete numerical methods for a nonlinear elliptic-parabolic system, the time-dependent Joule heating problem, are introduced and analyzed. The equations are discretized in space by a standard finite element method, and in time by combinations of rational implicit and explicit multistep schemes. The schemes are linearly implicit in the sense that they require, at each time level, the solution of linear systems of equations. Optimal order error estimates are proved under the assumption of sufficiently regular solutions. AMS subject classification (2000) 65M30, 65M15, 35K60     

Karush-Kuhn-Tucker systems: regularity conditions,error bounds and a class of Newton-type methods

 We consider optimality systems of Karush-Kuhn-Tucker (KKT) type, which arise, for example, as primal-dual conditions characterizing solutions of optimization problems or variational inequalities. In particular, we discuss error bounds and Newton-type methods for such systems. An exhaustive comparison of various regularity conditions which arise in this context is given. We obtain a new error bound under an assumption which we show to be strictly weaker than assumptions previously used for KKT systems, such as quasi-regularity or semistability (equivalently, the R ₀-property). Error bounds are useful, among other things, for identifying active constraints and developing efficient local algorithms. We propose a family of local Newton-type algorithms. This family contains some known active-set Newton methods, as well as some new methods. Regularity conditions required for local superlinear convergence compare favorably with convergence conditions of nonsmooth Newton methods and sequential quadratic programming methods. Received: December 10, 2001 / Accepted: July 28, 2002 Published online: February 14, 2003 Key words. KKT system – regularity – error bound – active constraints – Newton method Mathematics Subject Classification (1991): 90C30, 65K05     

Fuzzy modelling through logic optimization

This study proposes a new logic-driven approach to the development of fuzzy models. We introduce a two-phase design process realizing adaptive logic processing in the form of structural and parametric optimization. By recognizing the fundamental links between binary (two-valued) and fuzzy (multi-valued) logic, effective structural learning is achieved through the use of well-established methods of Boolean minimization encountered in digital systems. This blueprint structure is then refined by adjusting connections of fuzzy neurons, helping to capture the numeric details of the target system’s behavior. The introduced structure along with the learning mechanisms helps achieve high accuracy and interpretability (transparency) of the resulting model.     

Active classification using belief functions and information gain maximization

Obtaining reliable estimates of the parameters of a probabilistic classification model is often a challenging problem because the amount of available training data is limited. In this paper, we present a classification approach based on belief functions that makes the uncertainty resulting from limited amounts of training data explicit and thereby improves classification performance. In addition, we model classification as an active information acquisition problem where features are sequentially selected by maximizing the expected information gain with respect to the current belief distribution, thus reducing uncertainty as quickly as possible. For this, we consider different measures of uncertainty for belief functions and provide efficient algorithms for computing them. As a result, only a small subset of features need to be extracted without negatively impacting the recognition rate. We evaluate our approach on an object recognition task where we compare different evidential and Bayesian methods for obtaining likelihoods from training data and we investigate the influence of different uncertainty measures on the feature selection process.     

Approximating Unknown Mappings: An Experimental Evaluation

Different methodologies have been introduced in recent years with the aim of approximating unknown functions. Basically, these methodologies are general frameworks for representing non-linear mappings from several input variables to several output variables. Research into this problem occurs in applied mathematics (multivariate function approximation), statistics (nonparametric multiple regression) and computer science (neural networks). However, since these methodologies have been proposed in different fields, most of the previous papers treat them in isolation, ignoring contributions in the other areas. In this paper we consider five well known approaches for function approximation. Specifically we target polynomial approximation, general additive models (Gam), local regression (Loess), multivariate additive regression splines (Mars) and artificial neural networks (Ann).Neural networks can be viewed as models of real systems, built by tuning parameters known as weights. In training the net, the problem is to find the weights that optimize its performance (i.e. to minimize the error over the training set). Although the most popular method for Ann training is back propagation, other optimization methods based on metaheuristics have recently been adapted to this problem, outperforming classical approaches. In this paper we propose a short term memory tabu search method, coupled with path relinking and BFGS (a gradient-based local NLP solver) to provide high quality solutions to this problem. The experimentation with 15 functions previously reported shows that a feed-forward neural network with one hidden layer, trained with our procedure, can compete with the best-known approximating methods. The experimental results also show the effectiveness of a new mechanism to avoid overfitting in neural network training.     

O(2)-lnvariant Hamiltonians on ℂ4

A normal-form theory and a group-theoretic classification for periodic solutions of O(2)-invariant Hamiltonians on ?⁴ is developed. The theory applies to Hamiltonian systems with an O(2) spatial symmetry that also have a linear-mode interaction. Our motivation is the classic (m, n) mode-interaction problem for capillary-gravity waves. It is well known that the addition of surface-tension effects to irrotational water waves results in a countable infinity of values of the surface-tension coefficient at which two traveling waves of differing wavelength travel at the same speed. However, recognizing the reflectional symmetry in space, the linearized problem is actually spanned by four traveling waves. In other words there is an O(2) symmetry in space. A classification theorem for group-invariant Hamiltonian systems, based on a listing of the isotropy subgroups and their fixed-point spaces, is used to show that there are between seven and fourteen classes of periodic solutions in O(2)-invariant Hamiltonian systems with a mode interaction. The results are used to interpret, from a group-theoretic viewpoint, the classic Wilton ripple.     

Fuzzy rule generation methods for high-level computer vision

In many decision making systems involving multiple sources, the decisions made may be considered as the result of a rule-based system in which the decision rules are usually enumerated by experts or generated by a learning process. In this paper, we discuss the various issues involved in the generation of fuzzy rules automatically from training data for high-level computer vision. Features are treated as linguistic variables that appear in the antecedent clauses of the rules. We present methods to generate the corresponding linguistic labels (values) and their membership functions. Rules are generated by constructing a minimal approximate fuzzy aggregation network and then training the network using gradient descent methods. Several examples are given.     

Maximal words connected with unimodal maps

Orbits under a unimodal function are commonly characterized by letters of the alphabet (L, C, R). We present an algorithm for recognizing maximality of a word and study the necessary extension of the space of admissible words.     

基于PCP特征的和弦识别研究与探讨

讨论了音乐识别领域中和弦的四种不同识别方法,给出了基于PCP特征的和弦识别算法.使用PCP作为和弦的特征作为输入送至隐马尔可夫模型中训练,利用Baum-Welch算法估计模型参数,通过Viterbi算法得到正确和弦.通过实验获得了76%的识别率,验证了该算法的可行性.     

层次分析法在身份鉴别方案选择中的应用

针对身份鉴别方案比较的问题,采用层次分析法AHP(Analytic HierarchyProcess)建立了身份鉴别方案比较的数学模型,定量分析了指纹识别、人脸识别、语音识别、虹膜识别、静脉红外识别和DNA识别六种生物识别技术在身份鉴别中的综合表现,并做出评价.评价结果对重要场合安全检查中人员身份鉴别方法的选择有一定的指导意义.     

Krylov Subspaces Associated with Higher-Order Linear Dynamical Systems

A standard approach to model reduction of large-scale higher-order linear dynamical systems is to rewrite the system as an equivalent first-order system and then employ Krylov-subspace techniques for model reduction of first-order systems. This paper presents some results about the structure of the block-Krylov subspaces induced by the matrices of such equivalent first-order formulations of higher-order systems. Two general classes of matrices, which exhibit the key structures of the matrices of first-order formulations of higher-order systems, are introduced. It is proved that for both classes, the block-Krylov subspaces induced by the matrices in these classes can be viewed as multiple copies of certain subspaces of the state space of the original higher-order system. AMS subject classification (2000) 65F30, 15A57, 65P99, 41A21     

Graph-theoretical approach to qualitative solvability of linear systems

The structure of sign-solvable and strongly sign-solvable systems is studied here by a refinement of the graph-theoretic approach first suggested by Maybee. Both sign-solvability and strong sign-solvability are characterized in terms of an associated digraph. The problem of recognizing sign-solvability is reduced to two subproblems: recognizing strong sign-solvability and recognizing sign-nonsingularity. Under fairly general conditions on the sign patterns of A, it is possible to determine all sign patterns for b which render the system Ax = b sign-solvable.