Fuzzy Systems and Knowledge Discovery

As a generalization of fuzzy sets, the concept of interval-valued fuzzy sets was introduced by Gorzalczany [Fuzzy Sets and Systems

21

(1987) 1]. In this paper, we shall extend the concept of “fuzzy inclusion”, introduced by Šostak [Supp. Rend. Circ. Mat. Palermo (Ser. II)

11

(1985) 89], to the interval-valued fuzzy setting and study its fundamental properties for some extent.

A fuzzy evaluation based multi-objective optimization model for reactive power optimization in power distribution networks is presented in this paper. The two objectives, reducing active power losses and improving voltage profiles, are evaluated by membership functions respectively, so that the objectives can be compared in a single scale. To facilitate the solving process, a compromised objective is formed by the weighted sum approach. The weights are decided according to the preferences and importance of the objectives. The reactive tabu search algorithm is employed to get global optimization solutions. Simulation results of a practical power distribution network, greatly improved voltage profiles and reduced power losses, demonstrated that the proposed method is effective.

In this note, we introduce the concept of cut set of interval-valued fuzzy set and discuss some properties of cut set of interval-valued fuzzy set, propose three decomposition theorems of interval-valued fuzzy set and investigate some properties of cut set of interval-valued fuzzy set and mapping

H

in detail. These works can be used in setting up the basic theory of interval-valued fuzzy set.

It is important nowadays to provide guidance for individuals or organizations to improve their knowledge according to their objectives, especially in the case of incomplete cognition. Based on grey system theory, a knowledge architecture which consists of grey elements including knowledge fields and knowledge units is built. The method to calculate the weightiness of each knowledge unit, with regard to the user’s objectives, is detailed. The knowledge possessed by the user is also evaluated with grey clustering method by whitenization weight function.

Universal logic has given 0-level universal conjunction operation, universal disjunction operation and the universal implication operation. We introduce a new kind of algebra system UBL algebra based on these operations. A general propositional calculus formal deductive system

$\mathcal{L}^{U}$

of universal logic based on UBL algebras is built up, and its completeness is proved.

In this paper, we mainly extend entropy and subsethood from intuitionistic fuzzy sets to general interval-valued intuitionistic fuzzy sets, propose a definition of entropy and subsethood , offer a function of entropy and construct a class of subsethood function. Then from discussing the relationship between entropy and subsethood, we know that while choosing the subsethood, we can get some kinds of function of entropy based on subsethood. Our work is also applicable to practical fields such as: neural networks, expert systems, and other.

This paper presents a general framework for the study of fuzzy rough sets in which constructive approach is used. In the approach, a pair of lower and upper general fuzzy rough approximation operator in the lattice

L

is defined. Furthermore, the entire property and connection between the set of logical operator and the set of its corresponding general fuzzy rough approximation operator are examined, and we prove that they are 1-1 mapping. In addition, the structural theorem of negator operator is given. At last, the decomposition and synthesize theorem of general fuzzy rough approximation operator are proved. That for how to promote general rough approximation operator to suitable general fuzzy rough approximation operator, that is, how to select logical operator, provides theory foundation.

Associative classification is a well-known technique which uses association rules to predict the class label for new data object. This model has been recently reported to achieve higher accuracy than traditional classification approaches like

C4.5

. In this paper, we propose a novel associative classification algorithm based on correlation analysis,

ACBCA

, which aims at extracting the

k

-best strong correlated positive and negative association rules directly from training set for classification, avoiding to appoint complex support and confidence threshold.

ACBCA

integrates the advantages of the previously proposed effective strategies as well as the new strategies presented in this paper. An extensive performance study reveals that the improvement of

ACBCA

outperform other associative classification approaches on accuracy.

An approach to identify data-driven interpretable and accurate fuzzy models is presented in this paper. Firstly, Gustafson-Kessel fuzzy clustering algorithm is used to identify initial fuzzy model, and cluster validity indices are adopted to determine the number of rules. Secondly, orthogonal least square method and similarity measure of fuzzy sets are utilized to reduce the initial fuzzy model and improve its interpretability. Thirdly, constraint Levenberg-Marquardt algorithm is used to optimize the reduced fuzzy model to improve its accuracy. The proposed approach is applied to PH neutralization process, and results show its validity.

This paper presents a new kind of Evolutionary Fuzzy System (EFS) based on the Least Squares (LS) method and a hybrid learning algorithm: Adaptive Evolutionary-programming and Particle-swarm-optimization (AEPPSO). The structure of the Extended Fuzzy Basis Function Network (EFBFN) is firstly proposed, and the LS method is used to design it with presetting the widths of the hidden units in EFBFN. Then, to enhance the performance of the obtained EFBFN ulteriorly, a novel learning algorithm based on least squares and the hybrid of evolutionary programming and particle swarm optimization (AEPPSO) is proposed, in which we use EPPSO to tune the parameters of the premise part in EFBFN, and the LS algorithm to decide the consequent parameters in it simultaneously. In the simulation part, the proposed method is employed to predict a chaotic time series. Comparisons with some typical fuzzy modeling methods and artificial neural networks are presented and discussed.

The paper defines a fuzzy time Petri net (FTPN) which adopts four fuzzy set theoretic functions of time called fuzzy timestamp, fuzzy enabling time, fuzzy occurrence time and fuzzy delay, to deal with temporal uncertainty of train group operation and we also present different firing strategies for the net to give prominence to key events. The application instance shows that the method based on FTPN can efficiently analyze trains converging time, the possibility of converging and train terminal time in adjustment of train operation plan. Compared with time interval method, this method has some outstanding characteristics such as accurate analysis, simple computation, system simplifying and convenience for system integrating.

This paper deals with interval regression analysis using support vector machine and quantile regression method. The algorithm consists of two phases – the identification of the main trend of the data and the interval regression based on acquired main trend. Using the principle of support vector machine the linear interval regression can be extended to the nonlinear interval regression. Numerical studies are then presented which indicate the performance of this algorithm.

In this paper, we investigate the multiple attribute decision making problems under fuzzy linguistic environment. We introduce the concept of trapezoid fuzzy linguistic variable and some operational laws of trapezoid fuzzy linguistic variables. We develop a similarity measure between two trapezoid fuzzy linguistic variables. Based on the similarity measure and the ideal point of attribute values, we develop an approach to ranking the decision alternatives in multiple attribute decision making with trapezoid fuzzy linguistic variables. We finally illustrate the developed approach with a practical example.

Passenger satisfaction index (PSI) is one of the most important indexes in comprehensive evaluation of management performance and service quality of passenger transport corporations. Based on the investigations in China, the authors introduced the notion and method for passenger group division and the concept of index weight matrix, and made successful application for passenger satisfaction evaluation. Index weight matrix developed by applying AHP and Delphi methods gives satisfactory results. The paper ends with examples of using a fuzzy inference system for passenger satisfaction evaluation in Beijing railway station.

Based on the Fuzzy mathematical principle, a fuzzy integral model on forecasting the cancerational tendency of hydatidiform mole is created. In this paper, attaching function, quantum standard, weight value of each factor, which causes disease, and the threshold value of fuzzy integral value are determined under condition that medical experts take part in. The detailed measures in this paper are taken as follows: First, each medical expert gives the score of the sub-factors of each factor based on their clinic experience and professional knowledge. Second, based on analyzing the feature of the scores given by medical experts, attaching functions are established using K power parabola larger type. Third, weight values are determined using method by the analytic hierarchy process[AHP] method. Finally, the relative information is obtained from the case histories of hydatidiform mole cases. Fuzzy integral value of each case is calculated and its threshold value is finally determined. Accurate rate of the fuzzy integral model(FIM) is greater than that of the maximum likelihood method (MLM) via diagnosing the history cases and for new cases, the diagnosis results of the FIM is in accordance with those of the medical experts.

This paper presents a method for consensus measures and adjusting inconsistency of linguistic preference relations in group decision-making (GDM). Several consensus measures are defined by comparing positions of alternatives between individual ordered vectors and a collective ordered vector. In such a way, the consensus situation is evaluated in each moment in a more realistic way. Feedback mechanism would be applied if the consensus degree of all experts does not reach the required consensus level. This feedback mechanism is simple and easy computing process to help experts change their opinions in order to obtain the consensus level. It is based on the use of individual linguistic preferences, collective and individual ordered vectors of alternatives. It is also based on the use of fuzzy majority of consensus, represented by means of a linguistic quantifier.

Fuzzy set theory has grown to become a major scientific domain collectively referred to as fuzzy systems. In this paper, the fuzzy variation coefficients programming and its application are discussed. The proposed fuzzy variation coefficients programming has found application in operations management and it will find various application in other areas.

Dubois and Prade defined an interval-valued expectation of fuzzy numbers, viewing them as consonant random sets. Fullér and Majlender then proposed an weighted possibility mean value, variance and covariance of fuzzy numbers, viewing them as weighted possibility distributions. In this paper, we define a new weighted possibilistic variance and covariance of fuzzy numbers based on Fullér and Majlenders’ notations. Some properties of these notations are obtained in a similar manner as in probability theory. We also consider the weighted possibilistic mean-variance model of portfolio selection and introduce the notations of the weighted possibilistic efficient portfolio and efficient frontier. Moreover, a simple example is presented to show the application of our results in security market.

In this paper, we focus on the fuzzy constraints linear programming. First we discuss the properties of an optimal solution vector and of an optimal value in the corresponding parametric programming, and propose a method to the critical values. Then we present a new algorithm to the fuzzy constraint linear programming by associating an object function with the optimal value of parametric programming.

The concepts of fuzzy ultra filters and fuzzy G-filters of MTL-algebras are introduced. Some examples are given and the following main results are proved: (1) a fuzzy filter of MTL-algebra is fuzzy ultra filter if and only if it is both a fuzzy prime filter and fuzzy Boolean filter; (2) a fuzzy filter of MTL-algebra is fuzzy Boolean filter if and only if it is both a fuzzy G-filter and fuzzy MV-filter.

It is proved that a pair of dual fuzzy rough approximation operators can induce a topological space if and only if the fuzzy relation is reflexive and transitive. The sufficient and necessary condition that a fuzzy interior (closure) operator derived from a fuzzy topological space can associate with a fuzzy reflexive and transitive relation such that the induced fuzzy lower (upper) approximation operator is the fuzzy interior (closure) operator is also examined.

One of the fundamental goals of artificial intelligence (AI) is to build artificially computer-based systems which make computer simulate, extend and expand human’s intelligence and empower computers to perform tasks which are routinely performed by human beings. Without effective reasoning mechanism, it is impossible to make computer think and reason. Research on reasoning is an interesting and meaningful problem. Case-based reasoning is a branch of reasoning research. There are a lot of research results and successful application on Case-based reasoning. How to index and retrieve similar cases is one of key issue in Case-based reasoning research hotspots in CBR research works. In addition, there is a lot of uncertainty information in daily work and everyday life. So how to deal with uncertainty information in Case-based Reasoning effectively becomes more and more important. In this paper, a new case indexing and retrieval model which can deal with both fuzzy information and accurate information is presented.

This paper proposes an online self-organizing identification algorithm for TSK fuzzy model. The structure of TSK fuzzy model is identified using distance. Parameters of the piecewise linear function consisting consequent part are obtained using recursive version of combined learning method of global and local learning. Both input and output spaces are considered in the proposed algorithm to identify the structure of the TSK fuzzy model. By processing clustering both in input and output space, outliers are excluded in clustering effectively. The proposed algorithm is non-sensitive to noise not by using data itself as cluster centers. The proposed algorithm can obtain a TSK fuzzy model through one pass. By using the proposed combined learning method, the estimated function can have high accuracy.

We propose a generation method of membership function for extracting features of brain tissues on images of Magnetic Resonance Imaging (MRI). This method is derived from histogram analysis to create a membership function. According to

a priori

knowledge given by the neuro-radiologist, such as the features of gray level of differentiate brain tissues in MR images, we detect the peak or valley features of the histogram of MRI brain images. Then we determine a transformation of the histogram by selecting the feature values to generate a fuzzy membership function that corresponds to one type of brain tissues. A function approximations process is used to build a continuous membership function. This proposed method is validated for extracting whiter matter (WM), gray matter (GM), cerebra spino fluid (CSF). It is evaluated also using simulated MR images with two different, T1-weighted, T2-weighted MRI sequences. The higher agreement with the reference fuzzy model has been discovered by kappa statistic.

As a kind of new uncertainty theory, the set pair analysis (SPA) researches certainties and uncertainties from the whole. The main idea of SPA is: (1) Every system is comprised of certainty knowledge and uncertainty knowledge; (2) Certainties and uncertainties are inter-related, inter-influenced, inter-restricted, and inter-transformed even under certain condition, in every system; (3) A computation formula (identity-discrepancy-contrary (IDC) connection degree formula) which can embody the above idea fully is used to depict uniformly all kinds of uncertainties such as fuzzy uncertainty, random uncertainty, indeterminate-known uncertainty, unknown and unexpected incident uncertainty, and uncertainty which is resulted from imperfective information. This paper introduces the concepts and provides the applications of IDC connection degree and IDC connection number in SPA.

Automatic Summarization is need of the era. Mathematics is an important tool of nonfigurative thinking. A mathematic model of automatic summarization is established and discussed in the paper. The model makes use of meta-knowledge to describe the composition of the summary and help to calculate the semantic distance between summary and source document. It is proposed that how to get meta-knowledge aggregate and their weight are the key problems in the model.

In this paper, the selection of a data set from a universal set is carried out using a fuzzy entropy function. According to the definition of fuzzy entropy, the fuzzy entropy function is proposed and that function is proved through definitions. The proposed fuzzy entropy function calculates the certainty or uncertainty value of a data set; hence we can choose the data set that satisfies certain bounds or references. Therefore a reliable data set can be obtained using the proposed fuzzy entropy function. With a simple example we verify that the proposed fuzzy entropy function selects the reliable data set.

This paper proposes the formal description of nondeterministic information system based on tolerance rough set theory, analyzes six cases of approximate information system, and gives the concept of strong and weak similarity. After defining tolerance rough set, combining the theories of FCA and expanding non-definable concept into non-definable attributes, non-definable objects and non-definable context, we present optimal algorithm of formal concept of approximation system. Really emulation has illustrated that the algorithm obtains a satisfied approximate concept and a shorter time complexity.

In this paper, we propose a novel classification algorithm, called geometrical probability covering (GPC) algorithm, to improve classification ability. On the basis of geometrical properties of data, the proposed algorithm first forms extended prototypes through computing means of any two prototypes in the same class. Then Gaussian kernel is employed for covering the geometrical structure of data and used as a local probability measurement. By computing the sum of the probabilities that a new sample to be classified to the set of prototypes and extended prototypes, the classified criterion based on the global probability measurement is achieved. The proposed GPC algorithm is simple but powerful, especially, when training samples are sparse and small size. Experiments on several databases show that the proposed algorithm is promising. Also, we explore other potential applications such as outlier removal with the proposed GPC algorithm.

Typical description logics are limited to dealing with crisp concepts. It is necessary to add fuzzy features to description logics for management of the fuzzy information. In this paper, we propose extended fuzzy ALCN to enable representation and reasoning for complex fuzzy information. We define syntax structure, semantic interpretation and reasoning problems of the extended fuzzy ALCN, and discuss the reasoning properties inexistent in typical description logics. We also design tableau algorithms of reasoning problems for extended fuzzy ALCN. The tableau algorithms are developed in the style of so-called constraint propagation method. Extended fuzzy ALCN is more expressive than the existing fuzzy description logics and present more wide fuzzy information.

This paper propose a topological logic model of approximate reasoning based on the type II topological logic

$\mathbb{C}^2_\mathcal{T}$

and the structure of matching function

$\mathcal{C}$

and

$\mathcal{C}$

-match neighborhood group. The type II topological algorithm of simple approximate reasoning and multiple approximate reasoning is given in type II topological logic

$\mathbb{C}^2_\mathcal{T}$

with matching function

$\mathcal{C}$

. We also propose the structure of type II regular topological logic

$\mathbb{C}^2_\mathcal{T}$

and regular matching function

$\mathcal{C}$

. The type II completeness and type II perfectness of knowledge base

K

is investigated in type II regular topological logic

$\mathbb{C}^2_\mathcal{T}$

with regular matching function

$\mathcal{C}$

.

We introduce the consistent topological structure and neighborhood structure into the logical framework for providing the logical foundation and logical normalization for the approximate reasoning. We present the concept of the

formulae mass

, the

knowledge mass

and the

approximating knowledge closure

of the knowledge library by means of topological closure. We obtain the fundamental framework of type-I topological logics. In this framework, we present the type-I topological algorithm of the simple approximate reasoning and multi-approximate reasoning. In the frameworks of type-I strong topological logics, we present the type-I topological algorithm of multidimensional approximate reasoning and multiple multidimensional approximate reasoning. We study the type-I completeness and type-I perfection of the knowledge library in the framework of topological logical frameworks. We construct the type-I

knowledge universe

and prove that the second class knowledge universe of type-I is coincident with the first class knowledge universe of type-I, therefore the type-I knowledge universe is stable. We construct a self-extensive type-I knowledge library and the type-I expert system. In this expert system, the new approximate knowledge acquired by the self-extensive type-I knowledge library

K

I

will not beyond the type-I approximate knowledge closure, (

K

0

)

− −

, of the initial knowledge library

K

0

. Therefore, the precision of all new acquired approximate knowledge of this automatic reasoning system will be controlled well by the type-I approximate knowledge closure (

K

0

)

− −

of the initial knowledge library

K

0

.

We introduce a property of set to represent vagueness without using truth value. It has gotten less attention in fuzzy set theory. We introduce it by analyzing a well-known philosophical argument by Gearth Evans. To interpret ‘

a

is a vague object’ as ‘the Axiom of Extensionality is violated for

a

’ allows us to represent a vague object in Evans’s sense, even within classical logic, and of course within fuzzy logic.

In this paper, we use the notion of equivalence degree of fuzzy data, by which we can mine the rules of fuzzy functional dependencies from the fuzzy relational databases. Following the rules of fuzzy functional dependencies, we can apply the frame of analogical reasoning to refine the imprecise query answer for the relational databases with imprecise information.

The subject of this work is to establish a mathematical framework that provide the basis and tool for uncertainty reasoning based on linguistic information. This paper focuses on a flexible and realistic approach, i.e., the use of linguistic terms, specially, the symbolic approach acts by direct computation on linguistic terms. An algebra model with linguistic terms, which is based on a logical algebraic structure, i.e., lattice implication algebra, is applied to represent imprecise information and deals with both comparable and incomparable linguistic terms (i.e., non-ordered linguistic terms). Within this framework, some inferential rules are analyzed and extended to deal with these kinds of lattice-valued linguistic information.

This paper investigates lot sizing production planning problem with fuzzy unit profits, fuzzy capacities and fuzzy demands. First, the fuzzy production planning problem is formulated as a credibility measure based fuzzy programming model. Second, the crisp equivalent model is derived when the fuzzy parameters are characterized by trapezoidal fuzzy numbers. Then a fuzzy simulation-based genetic algorithm is designed for solving the proposed fuzzy programming model as well as its crisp equivalent. Finally, a numerical example is provided for illustrating the effectiveness of algorithm.

Comparison of fuzzy variables is considered one of the most important and interesting topics in fuzzy theory and applications. This paper introduces the new concept of fuzzy dominance based on credibility distributions of fuzzy variables. Some basic properties of fuzzy dominance are investigated. As an illustration, the first order case of fuzzy dominance rule for typical triangular fuzzy variables is examined.

In this paper, capital budgeting problem with fuzzy decisions is formulated as fuzzy chance-constrained programming models. Then fuzzy simulation, neural network and genetic algorithm are integrated to produce a hybrid intelligent algorithm for solving the proposed models. Finally, numerical experiments are provided to illustrate the effectiveness of the hybrid intelligent algorithm.

The derivative problems from the classical shortest path problem (SPP) are becoming more and more important in real life[1]. The dissimilar shortest paths problem is a typical derivative problem. In Vehicles Navigation System(VNS),it is necessary to provide drivers alternative paths to select. Usually, the path selected is a dissimilar path to the jammed path. In fact, ”dissimilar” is fuzzy. Considering traffic and transportation networks in this paper, we put forward to the definition of dissimilar paths measure that takes into account the decision maker’s preference on both the road sections and the intersections. The minimum model is formulated in which not only the length of paths but also the paths dissimilar measure is considered. And a genetic algorithm also is designed. Finally, we calculate and analyze the dissimilar paths in the traffic network of the middle and east districts of Lanzhou city in P.R. of China by the method proposed in this paper.

Fuzzy random variable is a measurable map from a probability space to a collection of fuzzy variables. In this paper, we first present several new convergence concepts for sequences of fuzzy random variables, including convergence almost sure, uniform convergence, almost uniform convergence, convergence in mean chance, and convergence in mean chance distribution. Then, we discuss the criteria for convergence almost sure, almost uniform convergence, and convergence in mean chance. Finally, we deal with the relationship among various types of convergence.

In this paper, hybrid genetic-SPSA algorithm based on random fuzzy simulation is proposed for solving chance-constrained programming in random fuzzy decision-making systems by combining random fuzzy simulation, genetic algorithm (GA), and simultaneous perturbation stochastic approximation (SPSA). In the provided algorithm, random fuzzy simulation is designed to estimate the chance of a random fuzzy event and the optimistic value to a random fuzzy variable, GA is employed to search for the optimal solution in the entire space, and SPSA is used to improve the new chromosomes obtained by crossover and mutation operations at each generation in GA. At the end of this paper, an example is given to illustrate the effectiveness of the presented algorithm.

This paper discusses the age-dependent replacement policy, in which the interarrival lifetimes of components are characterized as random fuzzy variables. A random fuzzy expected value model is presented and shown how it can be applied to reduce the loss of system failures. To solve the proposed model, a simultaneous perturbation stochastic approximation (SPSA) algorithm based on random fuzzy simulation is developed to search the optimal solution. At the end of this paper, a numerical example is enumerated.

In this paper, a new kind of alternating renewal processes—fuzzy random alternating renewal processes—is devoted. A theorem on the limit value of the mean chance of the fuzzy random event “system is on at time

t

” is presented. The two degenerate cases of the theorem, stochastic and fuzzy cases, are also analyzed. The importance of the results lies in the fact that the relation between classical alternating renewal processes and fuzzy random alternating renewal processes is established.

In this paper, a two-person zero-sum game is considered, in which the payoffs are characterized as fuzzy variables. Based on possibility measure, credibility measure, and fuzzy expected value operator, three types of concept of minimax equilibrium strategies,

r

-possible minimax equilibrium strategy,

r

-credible minimax equilibrium strategy, and expected minimax equilibrium strategy, are defined. An iterative algorithm based on fuzzy simulation is designed to find the equilibrium strategies. Finally, a numerical example is provided to illustrate the effectiveness of the algorithm.

In this paper, we propose a novel improved algorithm for the rectangular decomposition technique for the purpose of performing fuzzy knowledge discovery from large scaled database in a dynamic environment. To demonstrate its effectiveness, we compare the proposed one which is based on the newly derived mathematical properties with those of other methods with respect to the classification rate, the number of rules, and complexity analysis.

XML is an emerging standard for Internet data representation and exchange. There are more and more XML documents without associated schema on the Web. An XML document without associated schema is called well-formed XML document. It is difficult to store and query well-formed XML documents. This paper proposes a new XML documents storage model based on model-mapping, namely XPEV. The unique feature of model-mapping-based storage model is that no XML schema information is required for XML data storage. Through XPEV’s three tables: Path table, Edge table and Value table, well-formed XML documents could be easily stored in relational databases. XPEV model can make full use of the index technology of relational databases, and give a better solution for querying and storing well-formed XML documents using relational databases.

We propose a new nearest neighbor clustering classification algorithm based on fuzzy-rough set theory (FRNNC). First, we make every training sample fuzzy-roughness and use edit nearest neighbor algorithm to remove training sample points in class boundary or overlapping regions, and then use Mountain Clustering method to select representative cluster center points, then Fuzzy-Rough Nearest neighbor algorithm (FRNN) is applied to classify the test data. The new algorithm is applied to hand gesture image recognition, the results show that it is more effective and performs better than other nearest neighbor methods.

Text Categorization is the process of automatically assigning predefined categories to free text documents. Although there have existed a large number of text classification algorithms, most of them are either inefficient or too complex. In this paper, we propose the concept of category memberships, which stand for the degrees that words belonging to categories. Based on category memberships, a simple but efficient algorithm is presented. To evaluate our new algorithm, we have conducted experiments using Newsgroup_18828 text collection to compare it with Naive Bayes and

k

-NN. Experimental results show that our algorithm outperforms Naive Bayes and

k

-NN if a suitable category membership function is adopted.

In this paper, an efficient integrated location algorithm based on fuzzy identification and data fusion is presented in order to carry put precision and reliability position estimation and improve location accuracy and efficiency. In addition to the selectivity advantage gained by combining different location parameters, the use of integrated location algorithm by data fusion may increase location integrity with which a robust and anti-interference result can be obtained.

In this paper, a method of automatically identifying topics for Web documents via a classification technique is proposed. Web documents tend to have unpredictable characteristics, i.e. differences in length, quality and authorship. Motivated by these fuzzy characteristics, we adopt the fuzzy association concept to classify the documents into some predefined categories or topics. The experimental results show that our approach yields higher classification accuracy compared to the vector space model.

Fuzzy rules are suitable for describing uncertain phenomena and natural for human understanding and they are, in general, efficient for classification. In addition, fuzzy rules allow us to effectively classify data having non-axis-parallel decision boundaries, which is difficult for the conventional attribute-based methods. In this paper, we propose an optimized fuzzy rule generation method for classification both in accuracy and comprehensibility (or rule complexity). We investigate the use of genetic algorithm to determine an optimal set of membership functions for quantitative data. In our method, for a given set of membership functions a fuzzy decision tree is constructed and its accuracy and rule complexity are evaluated, which are combined into the fitness function to be optimized. We have experimented our algorithm with several benchmark data sets. The experiment results show that our method is more efficient in performance and comprehensibility of rules compared with the existing methods including C4.5 and FID3.1 (Fuzzy ID3).

Previous work considering both test and misclassification costs rely on the assumption that the test cost and the misclassification cost must be defined on the same cost scale. However, it can be difficult to define the multiple costs on the same cost scale. In our previous work, a novel yet efficient approach for involving multiple cost scales is proposed. Specifically speaking, we first introduce a new test-sensitive decision tree with two kinds of cost scales, that minimizes the one kind of cost and control the other in a given specific budget. In this paper, a dynamic test strategy with known information utilization and global resource control is proposed to keep the minimization of overall target cost. Our work will be useful in many urgent diagnostic tasks involving target cost minimization and resource consumption for obtaining missing information.

A linear matrix inequality approach to designing accurate classifier with a compact T–S(Takagi–Sugeno) fuzzy-rule is proposed, in which all the elements of the T–S fuzzy classifier design problem have been moved in parameters of a LMI optimization problem. Two-step procedure is used to effectively design the T–S fuzzy classifier with many tuning parameters: antecedent part and consequent part design. Then two LMI optimization problems are formulated in both parts and solved efficiently by using interior-point method. Iris data is used to evaluate the performance of the proposed approach. From the simulation results, the proposed approach showed superior performance over other approaches.

This paper presents new pruning and learning methods for the fuzzy rule-based classifier. For the simplicity of the model structure, the unnecessary features for each fuzzy rule are eliminated through the iterative pruning algorithm. The quality of the feature is measured by the proposed correctness method, which is defined as the ratio of the fuzzy values for a set of the feature values on the decision region to one for all feature values. For the improvement of the classification performance, the parameters of the proposed classifier are adjusted by using the gradient descent method so that the misclassified feature vectors are correctly re-categorized. Finally, the fuzzy rule-based classifier is tested on two data sets and is found to demonstrate an excellent performance.

A fuzzy set theory based region merging approach is presented to tackle the issue of oversegmentation from the watershed algorithm, for achieving robust image segmentation. A novel hybrid similarity measure is proposed as the merging criterion, based on the region-based similarity and the edge-based similarity. Both similarities are obtained using the fuzzy set theory. To adaptively adjust the influential degree of each similarity to region merging, a simple but effective weighting scheme is employed with the weight varying as region merging proceeds. The proposed approach has been applied to various images, including gray-scale images and color images. Experimental results have demonstrated that the proposed approach produces quite robust segmentations.

In this paper we report the combination of the Live-Wire method with the region growing algorithm based on fuzzy affinity. First, we employed anisotropic diffusion filter to process the images which smoothed the images while keeping the edge, and then we confined the possible boundary in applying the Live-Wire method to the over-segmentation found by the region growing algorithm. The speed and the reliability of the segmentation of the Live-Wire method are greatly improved by such combination. This method has been used for CT and MR image segmentation. The results confirmed that our method is practical and accurate in the medical image segmentation.

A new robust clustering scheme based on fuzzy c-means is proposed and the concept of a fuzzy mega-cluster is introduced in this paper. The fuzzy mega-cluster is conceptually similar to the noise cluster, designed to group outliers in a separate cluster. This proposed scheme, called the mega-clustering algorithm is shown to be robust against outliers. Another interesting property is its ability to distinguish between true outliers and non-outliers (vectors that are neither part of any particular cluster nor can be considered true noise). Robustness is achieved by scaling down the fuzzy memberships, as generated by FCM so that the

infamous

unity constraint of FCM is relaxed with the intensity of scaling differing across datum. The mega-clustering algorithm is tested on noisy data sets from literature and the results presented.

We present a method for extracting arbitrarily shaped clusters buried in uniform noise data. The popular

k

-means algorithm is firstly fuzzified with addition of entropic terms to the objective function of data partitioning problem. This fuzzy clustering is then kernelized for adapting to the arbitrary shape of clusters. Finally, the Euclidean distance in this kernelized fuzzy clustering is modified to a robust one for avoiding the influence of noisy background data. This robust kernel fuzzy clustering method is shown to outperform every its predecessor: fuzzified

k

-means, robust fuzzified

k

-means and kernel fuzzified

k

-means algorithms.

A fuzzy

c

-means algorithm incorporating the notion of dominant colors and spatial homogeneity is proposed for the color clustering problem. The proposed algorithm extracts the most vivid and distinguishable colors, referred to as the dominant colors, and then used these colors as the initial centroids in the clustering calculations. This is achieved by introducing reference colors and defining a fuzzy membership model between a color point and each reference color. The objective function of the proposed algorithm incorporates the spatial homogeneity, which reflects the uniformity of a region. The homogeneity is quantified in terms of the variance and discontinuity of the spatial neighborhood around a color point. The effectiveness and reliability of the proposed method is demonstrated through various color clustering examples.

Many clustering methods have been proposed in data mining fields, but seldom were focused on the incremental databases. In this paper, we present an incremental algorithm-IFHC that is applicable in periodically incremental environment based on FHC[3]. Not only can FHC and IFHC dispose the data with numeric attributes, but with categorical attributes. Experiment shows that IFHC is faster and more efficient than FHC in update of databases.

An efficient approximation of the Gaussian Probability Density Function (GPDF) is proposed in this paper. The proposed algorithm, called the Gradient-Based FCM with Divergence Measure (GBFCM (DM)), employs the divergence measurement as its distance measure and utilizes the spatial characteristics of MPEG VBR video data for MPEG data classification problems. When compared with conventional clustering and classification algorithms such as the FCM and GBFCM, the proposed GBFCM(DM) successfully finds clusters and classifies the MPEG VBR data modelled by the 12-dimensional GPDFs.

Although many systems exist for automatic classification of faces according to their emotional expression, these systems do not explicitly estimate the strength of given expressions. This paper describes and empirically evaluates an algorithm capable of estimating the degree to which a face expresses a given emotion. The system first aligns and normalizes an input face image, then applies a filter bank of Gabor wavelets and reduces the data’s dimensionality via principal components analysis. Finally, an unsupervised Fuzzy-C-Mean clustering algorithm is employed recursively on the same set of data to find the best pair of principle components from the amount of alignment of the cluster centers on a straight line. The cluster memberships are then mapped to degrees of a facial expression (i.e. less Happy, moderately happy, and very happy). In a test on 54 previously unseen happy faces., we find an orderly mapping of faces to clusters as the subject’s face moves from a neutral to very happy emotional display. Similar results are observed on 78 previously unseen surprised faces.

C-means clustering is a popular technique to classify unlabeled data into dif-ferent categories. Hard c-means (HCM), fuzzy c-means (FCM) and rough c-means (RCM) were proposed for various applications. In this paper a fuzzy rough c-means algorithm (FRCM) is present, which integrates the advantage of fuzzy set theory and rough set theory. Each cluster is represented by a center, a crisp lower approximation and a fuzzy boundary. The Area of a lower approximation is controlled over a threshold T, which also influences the fuzziness of the final partition. The analysis shows the proposed FRCM achieves the trade-off between convergence and speed relative to HCM and FCM. FRCM will de-grade to HCM or FCM by changing the parameter T. One of the advantages of the proposed algorithm is that the membership of clustering results coincides with human’s perceptions, which makes the method has a potential application in understandable fuzzy information granulation.

Image segmentation is an important research topic in image processing and computer vision community. In this paper, a new unsupervised method for MR brain image segmentation is proposed based on fuzzy c-means (FCM) and fuzzy connectedness. FCM is a widely used unsupervised clustering algorithm for pattern recognition and image processing problems. However, FCM does not consider the spatial coherence of images and is sensitive to noise. On the other hand, fuzzy connectedness method has achieved good performance for medical image segmentation. However, in the computation of fuzzy connectedness, one needs to select seeds manually which is elaborative and time-consuming. Our new method used FCM as the first step to select salient seeded points and then applied fuzzy connectedness algorithm based on those seeds. Thus our method achieved unsupervised automatic segmentation for brain MR images. Experiments on simulated and real data sets proved it is effective and robust to noise.

Algorithm of improved Fuzzy C-Means (FCM) clustering with preprocessing is analyzed and validated in the case of readout segmentation of photochromic optical disks. Characteristic of the readout and its differential coefficient and other knowledge are considered in the method, which makes it more applicable than the traditional FCM algorithm. The crest and trough segments could be divided clearly and the rising and falling edges could be located properly with the improved-FCM-based readout segmentation, which makes RLL encoding/decoding applicable to photochromic optical disks and makes the storage density increased. Further discussion proves the consistency of the segmentation method with PRML, and the improved-FCM-based detection could be regarded as an extension of PRML detection.

A good query plan in p2p networks is crucial to increase the query performance. Optimization of the plan requires effective and suitable remote cost estimation of the candidate peers on the basis of the information concerning the candidates’ run-time cost model and online time. We propose a fuzzy reward model to evaluate a candidate peer’s online reliability relative to the query host, and utilize a real-time cost model to estimate the query execution time. The optimizer is based on the run-time information to generate an effective query plan.

Classical fuzzy c-means and its variants cannot get better effect when the characteristic of samples is not obvious, and these algorithms run easily into locally optimal solution. According to the drawbacks, a novel mercer kernel based fuzzy clustering self-adaptive algorithm(KFCSA) is presented. Mercer kernel method is used to map implicitly the input data into the high-dimensional feature space through the nonlinear transformation. A self-adaptive algorithm is proposed to decide the number of clusters, which is not given in advance, and it can be gotten automatically by a validity measure function. In addition, attribute reduction algorithm is used to decrease the numbers of attributes before high dimensional data are clustered. Finally, experiments indicate that KFCSA may get better performance.

In this paper, we propose an improved information retrieval model, where the integration of modification-words and head-words is introduced into the representation of user queries and the traditional vector space model. We show how to calculate the weights of combined terms in vectors. We also propose a new strategy to construct the thesaurus in a fuzzy way for query expansion. Through the developed information retrieval system, we can retrieve documents in a relatively narrow search space and meanwhile extend the coverage of the retrieval to the related documents that do not necessarily contain the same terms as the given query. Experiments for testing the retrieval effectiveness have been implemented by using benchmark corpora. Experimental results show that the improved information retrieval system is capable of improving the retrieval performance both in precision and recall rates.

A new salient point extraction method from Discrete Cosine Transformation (DCT) compressed domain for content-based image retrieval is proposed in this paper. Using a few significant DCT coefficients, we provide a robust self-adaptive salient point extraction algorithm, and based on salient points, we extract 13 rotation-, translation- and scale-invariant moments as the image shape features for retrieval. Our system reduces the amount of data to be processed and only needs to do partial entropy decoding and partial de-qualification. Therefore, our proposed scheme can accelerate the work of image retrieval. The experimental results also demonstrate it improves performance both in retrieval efficiency and effectiveness.

Web documents that are either partially or completely duplicated in content are easily found on the Internet these days. Not only these documents create redundant information on the Web, which take longer to filter unique information and cause additional storage space, but they also degrade the efficiency of Web information retrieval. In this paper, we present a sentence-based copy detection approach on Web documents, which determines the existence of overlapped portions of any two given Web documents and graphically displays the locations of (semantically the) same sentences detected in the documents. Two sentences are treated as either the same or different according to the degree of similarity of the sentences computed by using either the

three least-frequent 4-gram

approach or the

fuzzy-set information retrieval

(

IR

) approach. Experimental results show that the fuzzy-set IR approach outperforms the three least-frequent 4-gram approach in our copy detection approach, which handles wide range of documents in different subject areas and does not require static word lists.

In document retrieval, expanding query with words that are semantically related or frequently co-occur can get good performance. In Chinese question answering system, in order to improve answer-document retrieval precision, query expansion is also necessary. Aiming at the specialty of Chinese question answering system, a method of query expansion based on related words for specific question types and synonym in HowNet is proposed. A computing method of similarity between questions and documents based on minimal matching span is presented. This method is based on vector space model, and also fully considers the position information of query words and query expansion words in the documents. Finally, the experiment results show that the effect of expanding query makes better than unexpanded one.

We present a new retrieval method based on multiple-Bernoulli model and multinomial model in this paper. We use the multiple-Bernoulli model and multinomial model to estimate the term probabilities by importing the conjugate prior and the term frequencies, and use Dirchlet method to smooth the models for solving the ”zero probability” problem of the language model.

The goal of information retrieval (IR) is to identify documents which best satisfy users’ information need. The task of formulating an effective query is difficult in the sense that it requires users to predict the keywords that will appear in the desired documents. In our study we proposed a method of query refinement by combining candidate keywords with query operators. The method uses the concept

Prime Keyword Set

, which is a subset of whole keywords and obtained by global analysis of the target database. Considering user’s intension we generate rational size of candidates by local analysis based on several specified principles. The experiments are conducted to confirm the effectiveness and efficiency of our proposed method. Moreover, as an extension of our approach an online system is implemented to investigate the feasibility.

In this paper, an Adaptive and Active Computing Paradigm (AACP) for personalized information service in heterogeneous environment is proposed to provide user-centered, push-based high quality information service timely in a proper way, the motivation of which is generalized as R4 Service: the Right information at the Right time in the Right way to the Right person, upon which formalized algorithms of adaptive user profile management, incremental information retrieval, information filtering, and active delivery mechanism are discussed in details. The AACP paradigm serves users in a push-based, event-driven, interest-related, adaptive and active information service mode, which is useful and promising for long-term user to gain fresh information instead of polling from kinds of information sources. Performance evaluations based on the AACP retrieval system that we have fully implemented manifest the proposed schema is effective, stable, feasible for adaptive and active information service in distributed heterogeneous environment.

The traditional algorithms of mining association rules, such as

Apriori

, often suffered from the bottleneck of itemset generation because the database is too large or the threshold of minimum support is not suitable. Furthermore, the traditional methods often treated each item evenly. It resulted in some problems. In this paper, a new algorithm to solve the above problems is proposed. The approach is to replace the database with the base set based on some seed items and assign weights to each item in the base set. Experiments on performance study will prove the superiority of the new algorithm.

The key technology in comparison-shopping is the online goods information extraction. Based on DOM, the information extraction with two-stage working pattern and the conception of page information unit have been proposed after a large number of sample pages testing. PIU is extracted and categorized by the classifying algorithm, and information is extracted from PIU. It is implemented that the key information of online goods is extracted based on the above-mentioned information extraction algorithm. It shows that the algorithm is steady and has higher Recall and Precision rate with the sample page testing.

Content-based image retrieval (CBIR) and semantic-based image retrieval (SBIR) have attracted great research attentions. However, they all have disadvantages. This paper proposes a retrieval method trying to overcome them. To achieve this we first introduce an approach of rough set-based low-level features selection. We propose an approach of feedback-based semantic-level features annotation. We also introduce a corresponding computing technology of similarity. We then build a model of combination of these two kinds of methods. Experimental results show that our method is more user-adaptive, and can achieve better performance compared with another retrieval method which is only based on low-level features.

An intelligent detect system to recognition unknown computer virus is proposed. Using the method based on fuzzy pattern recognition algorithm, a malicious executable code detection network model is designed also. This model target at Win32 binary viruses on Intel IA32 architectures. It could detect known and unknown malicious code by analyzing their behavior. We gathered 423 benign and 209 malicious executable programs that are in the Windows Portable Executable (PE) format as dataset for experiment . After extracting the most relevant API calls as feature, the fuzzy pattern recognition algorithm to detect computer virus was evaluated.

Risk in technological innovation is one of the important factors that hold enterprises from launching technological innovation. What cause the technological innovation risks is very complicated, and traditional methods of risk discernment can only draw general estimate on the risks. But enterprises need to understand the concrete links that cause technological innovation risks. For this reason, this paper puts forward a novel method of risk discernment in technological innovation, combining technological path graph with variable weight fuzzy evaluation, in order to clearerly, more accurately find the positions, in which technological innovation risks may take place, and evaluate the risks. Finally, the paper has verified the dependability of this method experimentally.

The prediction of epileptic seizures is a very attractive issue for all patients suffering from epilepsy in EEG (electroencephalograph) signals. It can assist to develop an intervention system to control / prevent upcoming seizures and change the current treatment method of epilepsy. This paper describes a new method based on wavelet transform and fuzzy similarity measurement to predict the seizures by using EEG signals. One part of the method is to calculate the energy and entropy of EEG data at the different scale; another part of this method is to calculate the similarity between the features set of the reference segment and the test segment using fuzzy measure. The test results of real rats show this method detect temporal dynamic changes prior to a seizure in real time.

This paper presents a multicriteria analysis (MA) approach to solve the problem of the optimal use of reserved land for agriculture involving multiple criteria and subjective assessments. Linguistic terms approximated by fuzzy numbers are used to adequately model the subjectiveness and imprecision of the decision making process. The degree of similarity between fuzzy numbers is used to calculate the overall performance index for each alternative across all criteria based on the concept of fuzzy ideal solution. As a result, the unreliable and often computationally demanding process of comparing fuzzy utilities usually required in fuzzy MA is avoided, and effective decisions can be made. A case study in Shanghai, China is presented that shows the fuzzy MA approach developed is efficient in computation, simple and comprehensible in concept, and practical in solving this kind of problems.

Studies on multi-group multi-criteria decision making problems for oil spill contingency management are in their infancy. This paper presents a second order fuzzy comprehensive evaluation (FCE) model to resolve decision-making problem in the area of contingency management after environmental disasters. To assess the performance of different oil combat strategies FCE allows the utilization of lexical information, consideration of ecological and socio-economic criteria and involvement of a variety of stakeholders. On the other hand, the approach can be validated by using internal and external checks, which refer to sensitivity tests regarding its internal setups and comparisons with other methods, respectively. Through a case study based on the Pallas oil spill occurred in German North-Sea 1998, it is demonstrated that this approach has wide application potential in the field of integrated coastal zone management.

Face recognition based only on the visual spectrum is not accurate or robust enough to be used in uncontrolled environments. This paper describes a fusion of visible and infrared (IR) imagery for face recognition. In this paper, a scheme based on membership function and fuzzy integral is proposed to fuse information from the two modalities. Recognition rate is used to evaluate the fusion scheme. Experimental results show the scheme improves recognition performance substantially.

This paper presents a novel group decision making method for integrating outcome preferences in the first-level hypergame models where each player correctly perceives the strategy set, but perceives possibly different outcome preferences of the opponent players. To get more correct preferences information in hypergame models, each player is often consisted of a group of decision makers who can give their perception about opponent players’ preferences respectively. In the face of opponent players’ different linguistic preferences relations over outcome space perceived by different decision makers, a group fuzzy preferences relation is first accurately computed using standard fuzzy arithmetic operations. Concept of consensus winner is then introduced to decide the crisp outcome preference vectors. A numerical example is provided at the end to illustrate the method.

In this paper, we study the multiple attribute group decision making (MAGDM) problems, in which the information about the attribute weights and the expert weights are interval numbers, and the attribute values take the form of uncertain linguistic information. We introduce some operational laws of uncertain linguistic variables and a formula for comparing two uncertain linguistic variables, and propose a new aggregation operator called interval aggregation (IA) operator. Based on the IA operatorand the formula for the comparison between two uncertain linguistic variables, we develop a method for MAGDM with uncertain linguistic information. Finally, an illustrative example is given to verify the developed method.

In this paper, we present a new prioritized information fusion method for handling fuzzy information retrieval problems. We also present a new center-of-gravity method for ranking generalized fuzzy numbers. Furthermore, we also extend the proposed prioritized information fusion method for handling fuzzy information retrieval problems in the generalized fuzzy number environment, where generalized fuzzy numbers are used to represent the degrees of strength with which documents satisfy particular criteria.

We propose a method of multiple context fusion based robust face detection scheme. It takes advantage of multiple contexts by combining color, illumination (brightness and light direction), spectral composition(texture) for environment awareness. It allows the object detection scheme can react in a robust way against dynamically changing environment. Multiple context based face detection is attractive since it could accumulate face model by autonomous learning process for each environment context category. This approach can be easily used in searching for multiple scale faces by scaling up/down the input image with some factor. The proposed face detection using the multiple context fusion shows more stability under changing environments than other detection methods. We employ Fuzzy ART for the multiple context- awareness. The proposed face detection achieves the capacity of the high level attentive process by taking advantage of the context-awareness using the information from illumination, color, and texture. We achieve very encouraging experimental results, especially when operation environment varies dynamically.

A novel adaptive algorithm for tracking maneuvering targets is proposed in this paper. The algorithm is implemented with fuzzy filtering and unscented transformation. A fuzzy system allows the filter to tune the magnitude of maximum accelerations to adapt to different target maneuvers. Unscented transformation act as a method for calculating the statistics of a random vector. A bearing-only tracking scenario simulation results show the proposed algorithm has a robust advantage over a wide range of maneuvers.

A new multisensor image fusion algorithm based on fuzzy logic is proposed. The membership function and fuzzy rules of the new algorithm is defined using the Fuzzy Inference System (FIS) editor of fuzzy logic toolbox in Matlab 6.1. The new algorithm is applied to fuse Charge-Coupled Device (CCD) and Synthetic Aperture Rader (SAR) images. The fusion result is compared with some other fusion algorithms through some performance evaluation measures for the fusion effect, and the comparison results show that the new algorithm is effective.

A great deal of research has been devoted in recent years to the designing Fuzzy-Neural Networks (FNN) from input-output data. And some works were also done to analyze the performance of some methods from a rigorous mathematical point of view. In this paper, the approximation bound for the clustering method, which is employed to design the FNN with the Bell Membership Function, is established. The detailed formulas of the error bound between the nonlinear function to be approximated and the FNN system designed based on the input-output data are derived.

It is a new idea and approach to forecast Web traffic basing on Neuro-Fuzzy Method. The log files on a Web Server include many useful information about users. In this paper, by analyzing log files the forecasting model is proposed and the basic idea, structure and algorithm of this model are introduced. The Dynamic Clustering Method and Neural Network learning method are introduced also. Experimental results show that the proposed method is very helpful for improving the administration of Web Server and quality of service and forecasting the action of users.

A voting-mechanism-based fuzzy neural network system based on generalized class cover problem and particle swarm optimization is proposed in this paper. When constructing the network structure, a generalized class cover problem and an improved greedy algorithm are adopted to get the class covers with relatively even radii, which are used to partition fuzzy input space and extract fewer robust fuzzy IF-THEN rules. Meanwhile, a weighted Mamdani inference mechanism is proposed to improve the efficiency of the system output and a particle swarm optimization-based algorithm is used to refine the system parameters. Experimental results show that the system is feasible and effective.

By utilizing some of the important properties of wavelets like denoising, compression, multiresolution along with the concepts of fuzzy logic and neural network, a new self-constructing fuzzy wavelet neural networks (SCFWNN) using compensatory fuzzy operators are proposed for intelligent fault diagnosis. An on-line learning algorithm is applied to automatically construct the SCFWNN. There are no rules initially in the SCFWNN. They are created and adapted as on-line learning proceeds via simultaneous structure and parameter learning. The advantages of this learning algorithm are that it converges quickly and the obtained fuzzy rules are more precise. The proposed SCFWNN is much more powerful than either the neural network or the fuzzy system since it can incorporate the advantages of both. The results of simulation show that this SCFWNN method has the advantage of faster learning rate and higher diagnosing precision.

This paper presents a new field balancing method for flexible rotors, which is based on adaptive neuro-fuzzy inference system (ANFIS) and information fusion. Firstly, new method fully utilizes the information supplied from all proximity sensors by holospectral technique for enhancing the balancing efficiency and accuracy. Secondly, a fuzzy model is established to simulate the mapping relationship between vibration responses and balancing weights using the ANFIS. The inputs of ANFIS are the amplitudes and phases of integrated vibration responses, while the outputs are the mass and azimuth of balancing weights. The experimental results show that the fuzzy balancing model based on ANFIS can obtain satisfactory balancing result after a single trial run, and possesses prospects of application in field balancing.

In this paper, we propose and evaluate a novel recognition algorithm for container identifiers that effectively overcomes these difficulties and recognizes identifiers from container images captured in various environments. The proposed algorithm, first, extracts the area containing only the identifiers from container images by using CANNY masking and bi-directional histogram method. The extracted identifier area is binarized by the fuzzy binarization method newly proposed in this paper. Then a contour tracking method is applied to the binarized area in order to extract the container identifiers, which are the target for recognition. This paper also proposes an enhanced fuzzy RBF network that adapts the enhanced fuzzy ART network for the middle layer. This network is applied to the recognition of individual codes. The results of experiment for performance evaluation on the real container images showed that the proposed algorithm performs better for extraction and recognition of container identifiers compared to conventional algorithms.

Data mining is the exploration and analysis, by automatic or semiautomatic means, of large quantities of data in order to discover meaningful patterns and rules. Data mining consists of several tasks and each task uses a variety of methodologies. Some of these tasks are suited for a top-down method called hypothesis testing and others are suited for a bottom-up method called knowledge discovery. In this paper, we report our research procedures and results that concern and relate ozone concentration data in various factors and attributes. We use the general steps of directed knowledge discovery methodologies and intelligent modeling techniques. Next, we construct ozone concentration prediction system in order to reduce various adverse effects on human beings and life on the earth.

Realistic understanding and description of car following behaviour is fundamental in many applications of Intelligent Transportation Systems. Historical car following studies had been focused on car following behaviour measured under experiment settings, either at test track or on open road, mainly using statistical analysis. This might introduce errors when they were used to represent everyday driving behaviour because differences might exist between everyday and experiment behaviours, and intelligent data analysis might be necessary in order to identify subtle differences. This paper presents the results of an observation and analysis of driver’s car following behaviour on motorway. Car following behaviours were measured under normal driving conditions where drivers were free to follow any vehicles. A time-series database was then established. The data was analysed using neuro-fuzzy systems and driver car following behaviour was quantified using several dynamic behavioural indices, which were combinations of parameters of trained neuro-fuzzy systems. The results indicated that in normal driving conditions, car following was conducted in a ‘loose’ way in terms of close-loop coupling, and car following performance was slightly ‘worse’ in terms of tracking error, than in experiment settings.

A genetic algorithm (GA) based composite sliding mode fuzzy control (CSMFC) approach is proposed for the double-pendulum-type overhead crane (DPTOC). The overhead crane exhibits double-pendulum dynamics because of the large-mass hook and the payload volume. Its nonlinear dynamic model is built using Lagrangian method. Through defining a composite sliding mode function, the proposed control approach greatly reduces the complexity to design a controller for complex underactuated systems. The control system stability is analyzed for DPTOC. Real-valued GA is used to optimize the parameters of CSMFC to improve the performance of control system. Simulation results illustrate the complexity of DPTOC and the validity of proposed control algorithm.

This paper deals with a balanced model reduction for a class of nonlinear systems with integral quadratic constraints(IQC’s) using a T-S(Takagi-Sugeno) fuzzy approach. We define a generalized controllability Gramian and a generalized observability Gramian for a stable T-S fuzzy systems with IQC’s. We obtain a balanced state space realization using the generalized controllability and observability Gramians and obtain a reduced model by truncating not only states but also IQC’s from the balanced state space realization. We also present an upper bound of the approximation error. The generalized controllability Gramian and observability Gramian can be computed from solutions of linear matrix inequalities.

The Non-gyro inertial measurement unit (NGIMU) uses only accelerometers replacing gyroscopes to compute the motion of a moving body. In a NGIMU system, an inevitable accumulation error of navigation parameters is produced due to the existence of the dynamic noise of the accelerometer output. When designing an integrated navigation system, which is based on a proposed nine-configuration NGIMU and a single antenna Global Positioning System (GPS) by using the conventional Kalman filter (CKF), the filtering results are divergent because of the complicity of the system measurement noise. So a fuzzy logic adaptive Kalman filter (FLAKF) is applied in the design of NGIMU/GPS. The FLAKF optimizes the CKF by detecting the bias in the measurement and prevents the divergence of the CKF. A simulation case for estimating the position and the velocity is investigated by this approach. Results verify the feasibility of the FLAKF.

Based on the analysis of the vehicle dynamics control system, the longitudinal control of tracking for the platoon of two vehicles is discussed. A second-order-model for the longitudinal relative distance control between vehicles is presented, and the logic switch rule between acceleration and deceleration of the controlled vehicle is designed. Then the parameters auto-adjusting fuzzy-PID control method is used by adjusting the three parameters of PID to control the varying range of error for the longitudinal relative distance and relative velocity between the controlled vehicle and the navigation vehicle, in order to realized the longitudinal control of a vehicle. The simulation results shown that compared with fuzzy control, the parameters auto-adjusting fuzzy-PID control method decreases the overshoot, enhances the capacity of anti-disturbance, and has certain robustness. The contradictory between rapidity and small overshoot has been solved.

This article describes a modeling approach based on an operator’s knowledge without a mathematical model of the system, and the optimization of the controller. The system used in this experiment could not easily be modeled by mathematical methods and could not easily be controlled by conventional systems. The controller was designed based on input-output data, and optimized under a predefined performance criterion.

The essential problem of multi-channel echo cancellation is caused by the strong correlation of two-channel input signals and the methods of pre-processing are always used to decorrelate it and the decorrelation degree depends on nonlinear coefficient

α

. But in most research,

α

is constant. In real application, the cross correlation is varying and

α

should be adjusted with correlation. But there is not precise mathematical formula between them. In this paper, the proposed method applies fuzzy logic to choose

α

so that the communication quality and convergence performance can be assured on the premise of small addition of computation. Simulations also show the effect of validity method.

Two different fuzzy control approaches are proposed for a class of nonlinear systems with mismatched uncertainties, transformable to the strict-feedback form. A fuzzy logic system (FLS) is used as a universal approximator to approximate unstructured uncertain functions and the bounds of the reconstruction errors are estimated online. By employing special design techniques, the controller singularity problem is completely avoided for the two approaches. Furthermore, all the signals in the closed-loop systems are guaranteed to be semi-globally uniformly ultimately bounded and the outputs of the system are proved to converge to a small neighborhood of the desired trajectory. The control performance can be guaranteed by an appropriate choice of the design parameters. In addition, the proposed fuzzy controllers are highly structural and particularly suitable for parallel processing in the practical applications.

The purpose of this paper is to investigate the use of evolutionary fuzzy neural systems to aircraft automatic landing control and to make the automatic landing system more intelligent. Three intelligent aircraft automatic landing controllers are presented that use fuzzy-neural controller with BPTT algorithm, hybrid fuzzy-neural controller with adaptive control gains, and fuzzy-neural controller with particle swarm optimization, to improve the performance of conventional automatic landing system. Current flight control law is adopted in the intelligent controller design. Tracking performance and adaptive capability are demonstrated through software simulations.

Because some timing-constraints of a controller task may be not determined as a real-time system engineer thinks of, its scheduling with uncertain attributes can not be usually and simply dealt with according to classic manners used in real-time systems. The model of a controller task with fuzzy deadline and its scheduling are studied. The dedication concept and the scheduling policy of largest dedication first are proposed first. Simulation shows that the scheduling of controller tasks with fuzzy deadline can be implemented by using the proposed method, whilst the control performance cost gets guaranteed.

Resource Management and scheduling is the important components of the Grid. It efficiently maps jobs submitted by the user to available resources in grid environment. Most mechanism about the resource scheduling focus on the performance of communication through the network and the load of the services, or the cost the users pay for the service. The performance and the cost are not the total factors the users consider. Much more constraint contained in the requests may make the jobs uncompleted by the services. Our work will consider the preference that the users make to the cost and deadline of his job. Even none of services can fit all the conditions in the request; there is always a service with the most satisfaction for the user.

A LMIs (linear matrix inequalities) based

H

 ∞ 

robust fuzzy control approach to an ultra-high rise/high speed elevator in the presence of uncertainties is presented in this paper. The uncertain nonlinear systems are represented using Takage-Sugeno (T-S) fuzzy models. The proposed controllers, which are in the form of the so-called parallel distributed compensation (PDC), stabilize nonlinear systems and guarantee an induced

L

2

norm bound constraint on disturbance attenuation for all admissible uncertainties. Finally, simulation results show the realization of the

H

 ∞ 

robust fuzzy control.

In this paper, a method to construct of a stable dual-mode predictive controller of unknown nonlinear system using the fuzzy system as a predictive model is proposed. The dual-mode controller is designed to ensure the stability in this region. In the neighborhood of the origin, a linear feedback controller designed for the linearized system generates the control action. Outside this neighborhood, predictive controller based on the fuzzy model is applied to the real nonlinear system. This method yields a stable closed-loop system when is applied to nonlinear systems under some conditions.

This paper presents a novel fuzzy modeling strategy using the hybrid algorithm EPPSO based on the combination of Evolutionary Programming (EP) and Particle Swarm Optimization (PSO) for control of nonlinear dynamical systems. The EPPSO is used to automatically design fuzzy controllers for nonlinear dynamical systems. In the simulation part, one multi-input multi-output (MIMO) plant control problem is performed. The performance of the suggested method is compared to that of EP, PSO and HGAPSO in the fuzzy controllers design. Simulation results demonstrate the superiority of the proposed method.

This paper studies an intelligent digital control for nonlinear systems with multirate sampling. It is worth noting that the multirate control design is addressed for a given nonlinear system represented by Takagi–Sugeno (T–S) fuzzy models. The main features of the proposed method are that it is provided that the sufficient conditions for stabilization of the discrete-time T–S fuzzy system derived by the fast discretization method in the sense of Lyapunov stability criterion, which is can be formulated in the linear matrix inequalities (LMIs).

As an assistant tool for human beings, humanoid robot is expected to cooperate with people to do certain jobs. Therefore, it must have high intelligence to adapt to common working condition. The objective of this paper is to propose an adaptive fuzzy logic control (FLC) method to improve system adaptability and stability, which can adjust hip and ankle joint based on sensor information. Furthermore, it can real time adjust controller parameters to improve FLC performance. Based on sensor information, humanoid robot can get environment and inherent situation and use the adaptive-FLC to realize stable locomotion. The effectiveness of the proposed method is shown with simulations based on the parameters of the “IHR-1” humanoid robot.

Computational intelligent techniques, e.g., neural networks, fuzzy systems, neuro-fuzzy systems, and evolutionary algorithms have been successfully applied for many engineering problems. These methods have been used for solving control problems in packet switching network architectures. The introduction of active networking adds a high degree of flexibility in customizing the network infrastructure and introduces new functionality. Therefore, there is a clear need for investigating both the applicability of computational intelligence techniques in this new networking environment, as well as the provisions of active networking technology that computational intelligence techniques can exploit for improved operation. We report on the characteristics of these technologies, their synergy and on outline recent efforts in the design of a computational intelligence toolkit and its application to routing on a novel active networking environment.

This paper deals with fuel-efficient maneuvers for constellation initialization. A new orbital controller based on fuzzy logic is proposed. It is composed of two level controllers: the high level controller is a fuzzy logic based planner, which resolves the conflicting constraints induced from the control efficiency and the control accuracy, while the base level controller is the well-known Folta-Quinn algorithm. The analysis and simulation studies indicate that the algorithm is very effective in reducing fuel cost for constellation formation capturing control.

In order to intercept the high-speed maneuverable targets in three-dimensional space, a terminal guidance law based on fuzzy logic systems is presented. After constructing the model of the relative movement between target and interceptor, guidance knowledge base which including fuzzy data and rules is obtained according to the trajectory performance. On the other hand, considering the time-variant and nonlinear factors in the thrust vector control system, the interceptor’s mass is identified in real time. By using the learning algorithms, the logic rules are also revised correspondingly to improve the fuzzy performance index. Simulation results show that this method can implement efficiently the precise guidance of the interceptor as well as preferable robust stability.

Relaxed linear matrix inequalities (LMIs) conditions for fuzzy observer-based controller design are proposed based on a kind of improved T-S fuzzy model structure. The improved structure included the original T-S fuzzy model and enough large bandwidth pre- and post-filters. By this structure fuzzy observer-based controller design can be transformed into LMIs optimization problem. Compared with earlier results, it includes the less number of LMIs that equals the number of fuzzy rules plus one positive definition constraint of Lyapunov function. Therefore, it provides us with less conservative results for fuzzy observer-based controller design. Finally, a numerical example is demonstrated to show the efficiency of proposed method.

Conventional virtual coupling is designed mainly for stabilizing the virtual environment (VE) and it thus may have poor performances. This paper proposes a novel adaptive virtual coupling design approach for haptic display in passive or time-delayed non-passive virtual environment. According to the performance errors, the virtual coupling can be adaptively tuned through some fuzzy logic based law. The designed haptic controller can improve the “operating feel” in virtual environments, while the system’s stability condition can still be satisfied. Experimental results demonstrate the effectiveness of this novel virtual coupling design approach.

A fuzzy model representation for describing the linguistic model of the object to be controlled in a control system is prompted. With the linguistic model of controlled object or process to be controlled, we can construct a close loop system representation. Consequently, we can discuss the system appearance with the assistance of the linguistic model as we do using a mathematic model in a conventional control system. In this paper, we discuss the describing ability of a fuzzy model and give a formal representation method for describing a fuzzy model. The combine method for a fuzzy system constructed by multiple fuzzy models is also discussed based on the controller model and the linguistic model of controlled object.

The novel fuzzy sliding mode control problem is presented for a class of uncertain nonlinear systems. The Takagi-Sugeno (T-S) fuzzy model is employed to represent a class of complex uncertain nonlinear system. A virtual state feedback technology is proposed to design the sliding mode plane. Based on Lyapunov stability theory, sufficient conditions for design of the fuzzy sliding model control are given. Design of the sliding mode controller based on reaching law concept is developed, which to ensure system trajectories from any initial states asymptotically convergent to sliding mode plane. The global asymptotic stability is guaranteed. A numerical example with simulation results is given to illustrate the effectiveness of the proposed method.

A new fuzzy control method for nonlinear pipeline system is discussed in this paper. The nonlinear dynamics of pipeline system is composed by two gravity-flow tanks, and are described by Takagi-Sugeno (T-S) fuzzy model. The controller design is based on overall stability, and is carried out via the parallel distributed compensation (PDC) scheme. To obtain better output dynamic performance, a given bounds is introduced to the output of nonlinear systems. Moreover, by means of linear matrix inequality (LMI) technique, it is shown that the existence of such constrained control system can be transformed into the feasibility of a convex optimization problem. Finally, by applying the designed controller, the simulation results demonstrate the efficiency.

In this paper a strategy of grading fuzzy sliding mode control (FSMC) applied in the AC servo system is presented. It combines the fuzzy logic and the method of sliding mode control, which can reduce the chattering without decreasing the system robustness. At the same time, the exponent approaching control is added by grading. The control strategy makes the response of the system quick and no overshoot. It is simulated to demonstrate the feasible of the proposed method by MATLAB6.5 and the good control effect is received.

The proposed controller in this paper, which combines the capability of fuzzy logic with the robustness of sliding mode controller, presents prevailing results with its adaptive architecture and proves to overcome the global stability problem of the control of nonlinear systems. Effectiveness of the controller and the performance comparison are demonstrated with chosen control techniques including PID and PD type self-tuning fuzzy controller on a quarter car model which consists of component-wise nonlinearities.

A new fuzzy modeling method using Multi-population Cooperative Particle Swarm Optimizer (MCPSO) for identification and control of nonlinear dynamic systems is presented in this paper. In MCPSO, the population consists of one master swarm and several slave swarms. The slave swarms executeParticle Swarm Optimization (PSO) or its variants independently to maintain the diversity of particles, while the particles in the master swarm enhance themselves based on their own knowledge and also the knowledge of the particles in the slave swarms. The MCPSO is used to automatic design of fuzzy identifier and fuzzy controller for nonlinear dynamic systems. The proposed algorithm (MCPSO) is shown to outperform PSO and some other methods in identifying and controlling dynamic systems.

This paper proposes computational intelligence for a perceptual system of a partner robot. The robot requires the capability of visual perception to interact with a human. Basically, a robot should perform moving object extraction, clustering, and classification for visual perception used in the interaction with human. In this paper, we propose a total system for human clustering for a partner robot by using long-term memory, k-means, self-organizing map and fuzzy controller is used for the motion output. The experimental results show that the partner robot can perform the human clustering.

This paper proposes a so-called fuzzy switching multiple (FSM) model which can achieve smooth switching when the control input at switching boundaries. Parallel distributed compensation scheme is employed to design the controller for the FSM. By utilizing fuzzy Lyapunov function, we derive the stabilization condition for closed-loop FSM systems. A design example illustrates the utility of the proposed approach.

Complex “lifelike” behaviors are composed of local interactions of individuals under fundamental rules of artificial life. In this paper, fundamental rules for cooperative group behaviors, “flocking” and “arrangement” of multiple autonomouse mobile robots are represented by a small number of fuzzy rules. Fuzzy rules in Sugeno type and their related parameters are automatically generated from clustering input-output data obtained from the algorithms for the group behaviors. Simulations demonstrate the fuzzy rules successfully realize group intelligence of mobile robots.

The paper addresses a study case in the frame of ground transportation aiming at improving service quality inside road tunnels. As part of a global project on vehicle automation which aim is to realize a reduced scale multi-sensor platoon of vehicles, a formal specification analysis is carried out and a UML-based design introduced taking into account different riding scenarios inside road tunnels besides a fuzzy control for longitudinal and lateral guidance of a caravan of vehicles. The proposed multimodel fuzzy controllers deal with the grappling and/or the unhooking of the automated train of vehicles for safe tunnels.

An analog Adaptive fuzzy logic controller is proposed. This controller is based on back-propagation algorithm, and designed for the use of on chip learning. This adaptive fuzzy logic controller was composed of an analog fuzzy logic controller and a learning circuit that realizes online learning mechanism. It can tune the consequent parameters automatically with the help of a direction signal. Hspice simulation of functional approximation experiment was held, showing that this controller has the on-chip learning capability.

A novel self-tuning fuzzy controller and its VLSI implementation are developed based on variable universe of discourse. This fuzzy controller is constructed by applying a contraction factor before each input of a conventional fuzzy controller, and a self-tuning gain factor after its output, while all the factors are adjusted with the input variables according to a simplified adaptive law. This fuzzy controller has some features: the contraction factors and output gain factor that improve the performance of the controller are based on variable universe of discourse; these factors are simplified for VLSI implementation; only the active rules are processed and the division in the operation of COA defuzzification is omitted by setting the denominator equal to 1. Results of Matlab – ActiveHDL co-simulation indicate that this self-tuning fuzzy controller works successfully in controlling a nonlinear system to track a reference trajectory.

A method to balance the communication among Multi-Agents in real time traffic synchronization is proposed in this research. The paper presents Air Traffic Flow Management (ATFM) problem and its synchronization property. For such a complex problem, combing grid computing with multi-agent coordination techniques to improve ATFM computational efficiency is the main objective of actual research. To demonstrate the developed model – ATFM in Grid Computing (ATFMGC), the grid architecture, the basic components and the relationship among them are described. At the same time, the function of agents (tactical planning agent etc.), their knowledge representation and inference processes are also discussed. As criteria to measure the effective to reduce quantity of the communication among agents and the delay of the flights, Standard of Balancing among Agents (SBA) is used in the analysis. The simulation shows the efficiency of the developed model and successful application in the case study.

Discovering frequent item sets is a key problem in data mining association rules. In this paper, there is a celerity association rules method based on data sort search. Using the plenitude and call terms of frequent item sets, the method efficiency can be improved greatly for the searching time won’t increase as the number of item set of the data does, moreover the data can be found by searching the database within 3 times. Using the change between the frequent item sets and standby item sets, the data celerity renew and the min-sup renew can be true.

In the Knowledge Economy era, trying to promote the whole competition advantage, the electronic businesses utilize information technology and internet to integrate the various kinds of application systems, database, and platform. It becomes common practice to construct an Enterprise Digital Content Portal (EDCP). To vary from minute to minute, coming to the problem of business environment is so difficult to integrate the complicated and huge amount information. By the way of collaboration of EDCP and the information of trade partners, it can provide the customers the real-time information that appearing with dynamic various timing. Through the combine of the information and procedure that between the enterprise and it’s business partner, it can use the assisting of information technology, improve the enterprise’s internal and external operational procedures, raise the transparency of information in the value chain, achieve the purpose that sharing information with different platform and language. In order to combine the different kinds of platform’s information between the different enterprises, we use Java technologies for web services in the construction and development of EDCP, use the Extended Markup Language, and the Web-based communication protocol, it communicates with other software system [1], accomplish the framework of the knowledge service platform that the enterprise deliver and communicate the internal and external information. In this paper, we propose the construction structure of the EDCP that using Web Services, we implement it into one case and utilize WebBench to be the analyzed tool that testing the efficiency when many people connect to the line at the same time, and prove the feasibility of this framework of the module. And we probe into the relevant literatures of collaboration, and make up the deficiency of relevant literatures in the past. Conduct to be the reference of the research that the enterprise and organization establish the relation of collaboration using EDCP in the future.

For a given product or object, predicting human emotions is very important in many business, scientific and engineering applications. There has been a significant amount of research work on the image-based analysis of human emotions in a number of research areas because human emotions are usually dependent on human vision. However, there has been little research on the computer image processing-based prediction, although such approach is naturally very appealing. In this paper, we discuss challenging issues in how to index images based on human emotions and present a heuristic approach to emotion-based image indexing. The effectiveness of image features such as colors, textures, and objects (or shapes) varies significantly depending on the types of emotion or image data. Therefore, we propose adaptive and selective techniques. With respect to six adverse pairs of emotions such as weak-strong, we evaluated the effectiveness of those techniques by applying them to the set of about 160 images in a commercial curtain pattern book obtained from the Dongdaemoon textile shopping mall in Seoul. Our preliminary experimental results showed that the proposed adaptive and selective strategies are effective and improve the accuracy of indexing significantly depending on the type of emotion.

To effectively reduce the computational loads during the optimization process, while maintaining the solution accuracy, a refined response surface method with design space transformation and refined RSM using sub-optimization for the regression model is proposed and implemented for the nose fairing design of a space launcher. Total drag is selected as the objective function, and the surface heat transfer, the fineness ratio, and the internal volume of the nose fairing are considered as design constraints. Sub-optimization for the design space transformation parameters and the iterative regression model construction technique are proposed in order to build response surface with high confidence level using minimum number of experiment points. The derived strategies are implemented to the nose fairing design optimization using the full Navier-Stokes equations. The result shows that an optimum nose fairing shape is obtained with four times less analysis calculations compared with the gradient-based optimization method, and demonstrates the efficiency of the refined response surface method and optimization strategies proposed in this study. The techniques can be directly applied to the multidisciplinary design and optimization problems with many design variables.

A MDO framework is a collaborative distributed computing environment that facilitates the integration of multi-disciplinary design efforts to achieve the global optimum result among local mutually-conflicting optimum results on heterogeneous platforms throughout the entire design process. The challenge for the MDO framework is to support the integration of legacy software and data, workflow management, heterogeneous computing, parallel computing and fault tolerance at the same time. In this paper, we present a Linda tuple space-based distributed computing framework optimized for MDO which is called MEDIC. In the design of MEDIC, we classify required technologies and propose an architecture in which those technologies can be independently implememnted at different layers. The Linda tuple space allows us to make the MEDIC architecture simple because it provides a flexible computing platform where various distributed and parallel computing models are easily implemented in the same way, multi-agents are easily supported, and effective fault tolerance techniques are available. A prototype system of MEDIC has been developed and applied for building an integrated design environment for super-high temperature vacuum furnaces called iFUD.

To search a pattern

P

in a text, index data structures such as suffix trees and suffix arrays are widely used. It is known that searching with suffix trees is faster than with suffix arrays in the aspect of time complexity. But recently, a few linear-time search algorithms for constant-size alphabet in suffix arrays have been suggested. One of such algorithms proposed by Sim et al. uses Burrows-Wheeler transform and takes

O

(|

P

|log|Σ|) time. But this algorithm needs too much space compared to Abouelhoda et al.’s algorithm to search a pattern.

In this paper we present an improved version for Sim et al.’s algorithm. It needs only 2

n

bytes at most if a given alphabet is sufficiently small.

Compared with the conventional ground rocket launching, air-launching has many advantages. However, comprehensive and integrated system design approach is required because the physical geometry of air launch vehicle is quite dependent on the installation limitation of the mother plane. Given mission objective is to launch 7.5kg nano-satellite to the target orbit of 700km x 700km using the mother plane, F-4E Phantom. The launching altitude and velocity are 12km, Mach number 1.5, respectively. As the propulsion system, a hybrid rocket engine is used for the first stage, and the solid rocket motors are used for the second and third stages. The total mass, length and diameter constraints of the rocket are imposed by the mother plane. The system design has been performed using the sequential optimization method. Gradient based SQP(Sequential Quadratic Programming) algorithm is employed. Analysis modules include mission analysis, staging, propulsion analysis, configuration, weight analysis, aerodynamics analysis and trajectory analysis. As a result of system optimization, a supersonic air launching rocket with total mass of 1272.61kg, total length of 6.43m, outer diameter of 0.60 m and the payload mass of 7.5kg has been successfully designed.

Recently, researches of microfluidics have been widely studied, because microfluidic systems have been used in several fields such as lab-on-a-chip in medicine, colloid thruster in aerospace, microhydrodynamic in engineering, etc. To handle and control the liquid in the microsystems, the hydrophobic and hydrophilic characteristics on a surface are very important properties. In this study, we performed numerical visualization to investigate the effect of surface wettability in microchannel on the flow characteristics. For the hydrophilic and hydrophobic surfaces arrangement, when the flow reaches the interface region, the meniscus shape changes from concave to convex and the velocity near the centre increases. We can present more efficient method to control the microflow in several micofluidic systems.

We propose a molecular modeling system based on web services. It visualizes three dimensional models of molecules and allows scientists observe and manipulate the molecular models “interactively” through the web. Scientists can examine, magnify, translate, rotate, combine and split the three dimensional molecular models. The real-time simulations are executed in order to validate the operations. We developed a distributed processing system for the real-time simulation.The proposed communication scheme reduces data traffics in the distributed processing system. The new job scheduling algorithm enhances the performance of the system. Thus, the scientists can interactively exercise molecular modeling procedures through the web. For the experiments, HIV-1 (Human Immunodeficiency Virus) was selected as a receptor and fifteen candidate materials were used as ligands. An experiment of measuring performance of the system showed that the proposed system was good enough to be used in molecular modeling on the web.

Effect of filter size of dynamic mixed model combined with a box filter on the prediction of passive scalar field has been investigated in complex flow. Unlike in the simple channel flow, the result shows that the model performance depends on the ratio of test to grid filter widths.

The development of CAM system for design and manufacturing of end mills becomes a key approach to save the time and reduce cost for end mills manufacturing. This paper presents the calculation and simulation of CNC machining end mill tools using on 5-axes CNC grinding machine tool. In this study the process of generation and simulation of grinding point data between the tool and the grinding wheels through the machined time are describes. Using input data of end mill geometry, wheels geometry, wheel setting, machine setting the end mill configuration and NC code for machining will be generated and visualized in 3 dimension before machining. The 3D visualizations of end mill manufacturing was generated by using OpenGL in C++.

The IP address lookup problem is to find the longest matching IP prefix from a routing table for a given IP address. In this paper we implemented and extended the results of [3] by incorporating the access frequencies of the target IP addresses. Experimental results showed that the number of memory access is reduced significantly.

This paper discusses a surface reconstruction method using the Delaunay triangulation algorithm. Surface reconstruction is used in various engineering applications to generate CAD model in reverse engineering, STL files for rapid prototyping and NC codes for CAM system from physical objects. The suggested method has two other components in addition to the triangulation: the weighted alpha shapes algorithm and the peel-off algorithm. The weighted alpha shapes algorithm is applied to restrict the growth of tetrahedra, where the weight is calculated based on the density of points. The peel-off algorithm is employed to enhance the reconstruction in detail. The results show that the increase in execution time due to the two additional processes is very small compared to the ordinary triangulation, which demonstrates that the proposed surface reconstruction method has great advantage in execution time for a large set of points.

This paper presents a clustering algorithm named HYBRID. HYBRID has two phases: in the first phase, a set of spherical

atom-clusters

with same size is generated, and in the second phase these atom-clusters are merged into a set of

molecule-clusters

. In the first phase, an incremental clustering method is applied to generate atom-clusters according to memory resources. In the second phase, using an edge expanding process, HYBRID can discover molecule-clusters with arbitrary size and shape. During the edge expanding process, HYBRID considers not only the distance between two atom-clusters, but also the closeness of their densities. Therefore HYBRID can eliminate the impact of outliers while discovering more isomorphic molecule-clusters. HYBRID has the following advantages: low time and space complexity, no requirement of users’ involvement to guide the clustering procedure, handling clusters with arbitrary size and shape, and the powerful ability to eliminate outliers.

The actual sound environment system exhibits various types of linear and non-linear characteristics, and it often contains an unknown structure. Furthermore, the observations in the sound environment are often contain fuzziness due to several causes. In this paper, a method for estimating the specific signal for acoustic environment systems with unknown structure and fuzzy observation is proposed by introducing a fuzzy probability theory and a system model of conditional probability type. The effectiveness of the proposed theoretical method is confirmed by applying it to the actual problem of psychological evaluation for the sound environment.

Discovery of interesting or frequently appearing time series patterns is one of the important tasks in various time series data mining applications. However, recent research criticized that discovering subsequence patterns in time series using clustering approaches is meaningless. It is due to the presence of trivial matched subsequences in the formation of the time series subsequences using sliding window method. The objective of this paper is to propose a threshold-free approach to improve the method for segmenting long stock time series into subsequences using sliding window. The proposed approach filters the trivial matched subsequences by changing Perceptually Important Point (PIP) detection and reduced the dimension by PIP identification.

In this paper, we propose an efficient algorithm which generates frequent itemsets by only one database scan. A frequent itemset is a set of common items that are included in at least as many transactions as a given minimum support. While scanning the database of transactions, our algorithm generates a table having 1-frequent items and a list of transactions per each 1-frequent item, and generates 2-frequent itemsets by using a hash technique.

k

(

k

≥3)-frequent itemsets can be simply found by checking whether for all (

k

–1)-frequent itemsets used to generate a

k

-candidate itemset, the number of common transactions in their lists is greater than or equal to the minimum support. The experimental analysis of our algorithm has shown that it can generate frequent itemsets more efficiently than FP-growth algorithm.

This paper begins with a new algorithm for computing time sequence data expansion distance on the time domain that, with a time complexity of O(n×m), solves the problem of retained similarity after the shifting and scaling of time sequence on the Y axis. After this, another algorithm is proposed for computing time sequence data expansion distance on frequency domain and searching similar subsequence in long time sequence, with a time complexity of merely O(n×fc), suitable for online implementation for its high efficiency, and adaptable to the extended definition of time sequence data expansion distance. An incremental DFT algorithm is also provided for time sequence data and linear weighted time sequence data, which allows dimension reduction on each window of a long sequence, simplifying the traditional O(n×m×fc) to O(n×fc).

MOLAP is a important application on multidimensional data warehouse. We often execute range queries on aggregate cube computed by pre-aggregate technique in MOLAP. For the cube with

d

dimensions, it can generate 2

d

cuboids. But in a high-dimensional cube, it might not be practical to build all these cuboids. In this paper, we propose a multi-dimensional hierarchical fragmentation of the fact table based on multiple dimension attributes and their dimension hierarchical encoding. This method partition the high dimensional data cube into shell mini-cubes. The proposed data allocation and processing model also supports parallel I/O and parallel processing as well as load balancing for disks and processors. We have compared the methods of shell mini-cubes with the other existed ones such as partial cube and full cube by experiment. The results show that the algorithms of mini-cubes proposed in this paper are more efficient than the other existed ones.

A popular solution to improving the speed and scalability of association rule mining is to do the algorithm on a random sample instead of the entire database. But it is at the expense of the accuracy of answers. In this paper, we present a sampling ensemble approach to improve the accuracy for a given sample size. Then, using Monte Carlo theory, we give an explanation for a sampling ensemble and obtain the theoretically low bound of sample size to ensure the feasibility and validity of an ensemble. And for learning the origination of the sample error and therefore giving theoretical guidance for obtaining more accurate answers, bias-variance decomposition is used in analyzing the sample error of an ensemble. According to theoretical analysis and real experiments, we conclude that sampling ensemble method can not only significantly improve the accuracy of answers, but also be a new means to solve the difficulty of determining appropriate sample size needed.

Distributed Data Mining (DDM) generally deals with the mining of data within a distributed framework such as local area and wide area networks. One strong case for DDM systems is the need to mine for patterns in very large databases. This requires mandatory partitioning or splitting of databases into smaller sets which can be mined locally over distributed hosts. Data Distribution implies communication costs associated with the need to combine the results from processing local databases. This paper considers the development of a DDM system on a cluster. In specific we approach the problem of data partitioning for data mining. We present a prototype system for DDM using a data partitioning mechanism based on Bayesian mixture modeling. Results from comparison with standard techniques show plausible support for our system and its applicability.

An effectual genetic coding is designed by constructing full-classification rule set. This coding results in full use of all kinds of excellent traits of genetic algorithm in data classification. The genetic algorithm is paralleled in process. So this leads to the improvement of classification and its ability to deal with great data. Some defects of current classifier algorithm are tided over by this algorithm. The analysis of experimental results is given to illustrate the effectiveness of this algorithm.

Knowledge reduction is one of the most important tasks in rough set theory, and most types of reductions in this area are based on complete information systems. However, many information systems are not complete in real world. Though several extended relations have been presented under incomplete information systems, not all reduction approaches to these extended models have been examined. Based on similarity relation, the similarity matrix and the upper/lower approximation reduction are defined under incomplete information systems. To present similarity relation with similarity matrix, the rough computational methods based on similarity relation are studied. The heuristic algorithms for non-decision and decision incomplete information systems based on similarity matrix are proposed, and the time complexity of algorithms is analyzed. Finally, an example is given to illustrate the validity of these algorithms presented.

This paper is devoted to the discussion of the relationship among some reduction approaches of information systems. It is proved that the distribution reduction and the entropy reduction are equivalent, and each distribute reduction is a

d

reduction. Furthermore, for consistent information systems, the distribution reduction, entropy reduction, maximum distribution reduction, distribute reduction, approximate reduction and

d

reduction are all equivalent.

Multi-attribute decision problems where the performances of the alternatives are random variables are considered in this paper. The suggested approach grades the probabilities of preference of one alternative over another with respect to the same attribute. Based on the graded probabilistic dominance relation, the pairwise comparison information table is defined. The global preferences of the decision maker can be seen as a rough binary relation. The present paper proposes to approximate this preference relation by means of the graded probabilistic dominance relation with respect to the subsets of attributes.

An incremental target recognition algorithm based on improved discernibility matrix in rough set theory is presented. Some comparable experiments have been completed in our “Information Fusion System for Communication Interception Information (IFS/CI

2

)”. The results of experimentation illuminate that the new algorithm is more efficient than the previous algorithm.

A relatively new area of research in centering on the phonetic encoding of information. This paper deals with the possible computer applications of the Sound Approach

©

English phonetic alphabet. The authors review some preliminary research into a few of the more promising approaches to the application of the processes of machine learning to this phonetic alphabet for computer spell-checking, computer speech recognition etc. Applying mathematical approaches to the development of a data-based phonetic spelling recognizer, and speech recognition technology used for language pronunciation training in which the speech recognizer allows a large margin of pronunciation accuracy, the authors delineate the parameters of the current research, and point the direction of both the continuation of the current project and future studies.

Multiple classifier systems have become a popular classification paradigm for strong generalization performance. Diversity measures play an important role in constructing and explaining multiple classifier systems. A diversity measure based on relation entropy is proposed in this paper. The entropy will increase with diversity in ensembles. We introduce a technique to build rough decision forests, which selectively combine some decision trees trained with multiple reducts of the original data based on the simple genetic algorithm. Experiments show that selective multiple classifier systems with genetic algorithms get greater entropy than those of the top-classifier systems. Accordingly, good performance is consistently derived from the GA based multiple classifier systems although accuracies of individuals are weak relative to top-classifier systems, which shows the proposed relation entropy is a consistent diversity measure for multiple classifier systems.

An efficient design method to improve the control performance of rough controller is presented in this paper. As the input-output data of the history process operation may not be enough informative, extra testing signals are used to excite the process to acquire sufficient data reflecting the control laws of the operator or the existing controller. Using data from the successive exciting tests or excellent operation by operators, the rules can be updated and enriched, which is helpful to improve the performance of the rough controller. The effectiveness of the proposed method is demonstrated through two simulation examples emulating PID control and Bang-Bang control, respectively.

At present, much more research in the field of soft sensor modeling is concerned. In the process of establishing soft sensor models, how to select the secondary variables is still an unresolved question. In this paper, rough set theory is used to select the secondary variables from the initial sample data. This method is used to build the soft sensor model to estimate the oxygen concentration in a regeneration tower and the good result is obtained.

The problem of real value attribute discretization can be converted into the reduct problem in the Rough Set Theory, which is NP-hard and can be solved by some heuristic algorithms. In this paper we show that the straightforward conversion is not scalable and propose a divide-and-conquer algorithm. This algorithm is fully scalable and can reduce the time complexity dramatically especially while integrated with the tournament discretization algorithm. Parallel versions of this algorithm can be easily written, and their complexity depends on the number of objects in each subtable rather than the number of objects in the initial decision table. There is a tradeoff between the time complexity and the quality of the discretization scheme obtained, and this tradeoff can be made through adjusting the number of subtables, or equivalently, the number of objects in each subtable. Experimental results confirm our analysis and indicate appropriate parameter setting.

Rough set theory (RST) can mine useful information from a large number of data and generate decision rules without prior knowledge. Support vector machines (SVMs) have good classification performances and good capabilities of fault-tolerance and generalization. To inherit the merits of both RST and SVMs, a hybrid classifier called rough set support vector machines (RS-SVMs) is proposed to recognize radar emitter signals in this paper. RST is used as preprocessing step to improve the performances of SVMs. A large number of experimental results show that RS-SVMs achieve lower recognition error rates than SVMs and RS-SVMs have stronger capabilities of classification and generalization than SVMs, especially when the number of training samples is small. RS-SVMs are superior to SVMs greatly.

Attribute reduction is one of the basic contents in decision table. And it has been proved that computing the optimal attribute reduction is NP-complete. A lot of algorithms for the optimal attribute reduction were proposed in consistent decision table. But most decision tables are inconsistent in fact. In this paper, the judgment theorem with respect to maximum distribution reduction is obtained and the significance of attributes is defined in decision table, from which a polynomial heuristic algorithm for the optimal maximum distribution reduction is proposed. Finally, the experimental results show that this algorithm is effective and efficient.

In the axiomatic approach of fuzzy rough set theory, fuzzy rough approximation operators are characterized by a set of axioms that guarantees the existence of certain types of fuzzy binary relations reproducing the operators. Thus axiomatic characterization of fuzzy rough approximation operators is an important aspect in the study of rough set theory. In this paper, the independence of axioms of generalized fuzzy rough approximation operators is investigated, and their minimal sets of axioms are presented.

Rough set (RS) theory is a mathematics tool for handling uncertain problem. It is helpful for KDD, but expensive consumption of time and unclear expression of result are the main problem in practical application. The extended concept lattice (ECL) jis a new form of concept lattice which is gotten by introducing equivalence intension into Galois concept lattice (GCL). The ECL is an efficient tool for data analysis and knowledge discovery in database (KDD). Both ECL and RS are based on equivalence class, so the relative between them exists. This paper describes the ECL first, then discusses the relation between the ECL and RS, and describes the implementation of rough set based on ECL.

This paper firstly compares rough sets theory with grey system theory, then the concept of grey sets is proposed and the grey degree of grey sets and the basic relationships and operations of grey sets are defined. Next, we set up the models of grey rough sets and wide grey rough sets as well as study their properties. Moreover, the definition of rough grey sets is given and their basic characters are investigated. Furthermore, the relations between rough grey sets and grey rough sets are researched. Finally, we come to the conclusion that it is possibly more effective to deal with some uncertain problems if the theories and methods of rough sets and grey systems are combined.