Search

The paper shows methods that could be suitably applied to manage uncertainties in space systems design and modeling. Attention is focused on the Interval Analysis, as one of the most promising methods. In this research two issues typical faced in space-related researchs have been treated: optimization and ordinary differential equations (ODEs) solving taking into account possible uncertainties included in the models; the first is generally related to space system design phases while the second deals with systems dynamics analysis. Because of lack of space this paper presents results from the first area only: more specifically, the optimization of a control system design to accomplish robust reconfiguration manoeuvre is achieved taking into account uncertainties on the technological parameters of the propulsion unit to obtain a robust design from the beginning of the study. Results turned out that the Interval Analysis is effective and efficient to handle these classes of uncertainties. [ABSTRACT FROM AUTHOR]

Different clustering algorithms are based on different similarity or distance measures (e.g. Euclidian distance, Minkowsky distance, Jackard coefficient, etc.). Jarvis-Patrick clustering method utilizes the number of the common neighbors of the k-nearest neighbors of objects to disclose the clusters. The main drawback of this algorithm is that its parameters determine a too crisp cutting criterion, hence it is difficult to determine a good parameter set. In this paper we give an extension of the similarity measure of the Jarvis-Patrick algorithm. This extension is carried out in the following two ways: (i) fuzzyfication of one of the parameters, and (ii) spreading of the scope of the other parameter. The suggested fuzzy similarity measure can be applied in various forms, in different clustering and visualization techniques (e.g. hierarchical clustering, MDS, VAT). In this paper we give some application examples to illustrate the efficiency of the use of the proposed fuzzy similarity measure in clustering. These examples show that the proposed fuzzy similarity measure based clustering techniques are able to detect clusters with different sizes, shapes and densities. It is also shown that the outliers are also detectable by the proposed measure. [ABSTRACT FROM AUTHOR]

In this paper we analyze the convergence of independent adaptive learners in repeated games. We show that, in this class of games, independent adaptive learners converge to pure Nash equilibria in self play, if they exist, and to a best response strategy against stationary opponents. We discuss the relation between our result and convergence results of adaptive play [1]. The importance of our result stems from the fact that, unlike adaptive play, no communication/action observability is assumed. We also relate this result to recent results on the convergence of weakened ficticious play processes for independent learners [2,3]. Finally we present experimental results illustrating the main ideas of the paper. [ABSTRACT FROM AUTHOR]

This paper describes a complete and efficient vision system developed for the robotic soccer team of the University of Aveiro, CAMBADA (Cooperative Autonomous Mobile roBots with Advanced Distributed Architecture). The system consists on a firewire camera mounted vertically on the top of the robots. A hyperbolic mirror placed above the camera reflects the 360 degrees of the field around the robot. The omnidirectional system is used to find the ball, the goals, detect the presence of obstacles and the white lines, used by our localization algorithm. In this paper we present a set of algorithms to extract efficiently the color information of the acquired images and, in a second phase, extract the information of all objects of interest. Our vision system architecture uses a distributed paradigm where the main tasks, namely image acquisition, color extraction, object detection and image visualization, are separated in several processes that can run at the same time. We developed an efficient color extraction algorithm based on lookup tables and a radial model for object detection. Our participation in the last national robotic contest, ROBOTICA 2007, where we have obtained the first place in the Medium Size League of robotic soccer, shows the effectiveness of our algorithms. Moreover, our experiments show that the system is fast and accurate having a maximum processing time independently of the robot position and the number of objects found in the field. [ABSTRACT FROM AUTHOR]

Research in the field of examination timetabling has developed in two directions. The first looks at applying various methodologies to induce examination timetables. The second takes an indirect approach to the problem and examines the generation of heuristics or combinations of heuristics, i.e. hyper-heuristics, to be used in the construction of examination timetables. The study presented in this paper focuses on the latter area. This paper presents a first attempt at using genetic programming for the evolution of hyper-heuristics for the uncapacitated examination timetabling problem. The system has been tested on 9 benchmark examination timetabling problems. Clash-free timetables were found for all 9 nine problems. Furthermore, the performance of the genetic programming system is comparable to, and in a number of cases has produced better quality timetables, than other search algorithms used to evolve hyper-heuristics for this set of problems. [ABSTRACT FROM AUTHOR]

This paper presents a pruning algorithm based on optimal brain surgeon (OBS) for general dynamic neural networks (GDNN). The pruning algorithm uses Hessian information and considers the order of time delay for saliency calculation. In GDNNs all layers have feedback connections with time delays to the same and to all other layers. The parameters are trained with the Levenberg-Marquardt (LM) algorithm. Therefore the Jacobian matrix is required. The Jacobian is calculated by real time recurrent learning (RTRL). As both LM and OBS need Hessian information, a rational implementation is suggested. [ABSTRACT FROM AUTHOR]

Pawlak introduced approximation spaces in his seminal work on rough sets more than two decades ago. In this paper, we show that approximation spaces are basic structures for knowledge discovery from multi-relational data. The utility of approximation spaces as fundamental objects constructed for concept approximation is emphasized. Examples of basic concepts are given throughout this paper to illustrate how approximation spaces can be beneficially used in many settings. The contribution of this paper is the presentation of an approximation space-based framework for doing research in various forms of knowledge discovery in multi relational data. [ABSTRACT FROM AUTHOR]

This paper extends the basic rough set formalism introduced by Pawlak [1] to a rule-based knowledge representation language, called Rough Datalog, where rough sets are represented by predicates and described by finite sets of rules. The rules allow us to express background knowledge involving rough concepts and to reason in such a knowledge base. The semantics of the new language is based on a four-valued logic, where in addition to the usual values True and False, we also have the values Boundary, representing uncertainty, and Unknown corresponding to the lack of information. The semantics of our language is based on a truth ordering different from the one used in the well-known Belnap logic [2, 3] and we show why Belnap logic does not properly reflect natural intuitions related to our approach. The declarative semantics and operational semantics of the language are described. Finally, the paper outlines a query language for reasoning about rough concepts. [ABSTRACT FROM AUTHOR]

Many monitoring and control computer systems contain a rule-based system as a part of them. Such a rule based system is used to determine which actions should be taken depending on the data collected from sensors. For both embedded and rule-based systems many different approaches have been proposed, but it is hardly possible to find a formalism that can cope with both of them. The paper deals with a problem of including decision tables into colour Petri net models. A few kinds of decision tables are considered and methods of transformation them into coloured Petri nets form called D-nets are presented. Both non-hierarchical and hierarchical D-nets are considered in the paper. [ABSTRACT FROM AUTHOR]

In this paper, the design, simulation, implementation and experiment of rough set processor are described. The experiment result shows that the proposed processor is ten times faster than PC, though the clock frequency is about 70 times lower. [ABSTRACT FROM AUTHOR]

Detection of natural selection at the molecular level is one of the crucial problems in contemporary population genetics. There exists a number of statistical tests designed for it, however, the interpretation of the outcomes is often obscure, because of the existence of factors like population growth, migration and recombination. In his earlier work, the author has proposed the multi-null methodology, and he applied it for four genes implicated in human familial cancer: ATM, RECQL, WRN and BLM. Because of high computational effort required for estimating critical values under nonclassical nulls, mentioned methodology is not appropriate for selection screening. In the current paper, the author presents novel, rough set based methodology, helpful in the interpretation of tests outcomes applied versus only classical nulls. This method does not require long-lasting simulations and, as it is shown in the paper, it gives reliable results. [ABSTRACT FROM AUTHOR]

In this paper a method of creating layers of feed-forward neural network that does not need to be learned is presented. Described approach is based on algorithms used in synthesis of logic circuits. Experimental results presented in the paper prove that this method may significantly decrease the time of learning process, increase generalization ability and decrease a probability of sticking in a local minimum. Further work and goals to achieve are also discussed. [ABSTRACT FROM AUTHOR]

Synonymy has been of high importance in information retrieval and automatic indexing. Recently, in the view of special needs for domain ontology building and maintenance, the problem returns with a higher demand. In the presented paper, we present a novel text mining approach to discovering synonyms or close meaning terms. The offered measures of closeness of terms (or their contexts) are expressed by means of data mining notions; namely, frequent termsets and association rules. The measures can be calculated by using data mining techniques, such as the well known Apriori algorithm. The approach is domain-independent and large-scale. It is, however, restricted to the recognition of parts of speech. In that sense the approach is language dependent, up to the language dependency of the parts of speech tagging process. The experimental results obtained with the approach are presented. [ABSTRACT FROM AUTHOR]

The identification of appropriate text tokens (words or sequences of words representing concepts) is one of the most important tasks of text preprocessing and may have great influence on the final results of text analysis. In our paper, we introduce a new approach to discovering compound nouns, including proper compound nouns. Our approach combines the data mining methods with shallow lexical analysis. We propose a simple pattern language for specifying grammatical patterns to be satisfied by extracted compound nouns. Our method requires annotating the words with part of speech tags, thus to this extent, it is language-dependent. Based on the data mining GSP algorithm, we propose T-GSP as its modification for extracting frequent text patterns, and in particular, frequent word sequences that satisfy given grammatical rules. The obtained sequences are regarded as candidates for compound nouns. The experiments have proven very high quality of the method. [ABSTRACT FROM AUTHOR]

As a result of advances in technology, large amounts of data can be collected and stored automatically. Significant development of the Internet and easier access to it have contributed to collecting large amounts of information about users' characteristics. Along with these changes, concerns about privacy of data have emerged. Several methods of preserving privacy for association rules mining have been proposed in literature: MASK scheme and its optimizations. This paper provides new solutions concerning efficiency for this scheme and considers different methods of distorting data using randomization techniques. Effectiveness of these solutions has been tested and presented in this paper. [ABSTRACT FROM AUTHOR]

Reduction in rough set theory is useful to compact given attributes of large-scale decision tables in data mining. In this paper a new method called grey-rough reduction is proposed for decision tables containing non-interval data and interval data complexly called grey-decision tables. First of all, a grey-rough approximation is introduced after summarized grey numbers, their operations and functions. Two sorts of reduction based on grey-rough sets, a basic approach and advanced approach are proposed with several illustrative examples. Three experiments, compatibility with the classical model, an application of the basic approach to decision-making and influence of the parameter in the advanced approach are shown. The advantages of the proposal are (1) it is compatible with the classical reduction model for non-interval data, (2) it is useful for complex decision tables and (3) it provides a possible reduction of attributes with a parameter by the advanced approach. [ABSTRACT FROM AUTHOR]

Non-standard logics departs from traditional logic mostly in extended views, on one hand syntactically related to logical operators, and on the other hand semantically related to truth values. Typical for these approaches is the remaining traditional view on 'sets and relations' and on terms based on signatures. Thus the cornerstones of the languages remain standard, and so does mostly the view on knowledge representation and reasoning using traditional substitution theories and unification styles. In previous papers we have dealt with particular problems such as generalizing terms and substitution, extending our views on sets and relations, and demonstrated the use of these non-standard language elements in various applications such as for fuzzy logic, generalized convergence spaces, rough sets and Kleene algebras. In this paper we provide an overview and summarized picture of what indeed happens when we drop the requirement for using traditional sets with relations and terms with equational settings. [ABSTRACT FROM AUTHOR]

The present paper is concerned with Rough Set Theory (RST) and Similarity Coverage Model (SCM) of category-based induction. It redefines basic concepts of RST in the light of SCM, and explains how RST may be seen as an elegant formal model of inductive reasoning. Furthermore, following SCM, we enrich RST by the concept of an ontology defined as a subset of the family of all definable sets. The paper also presents a model of inductive reasoning which is driven by recent works on RST and nearness-type structures. We show how approximation spaces can be characterised in terms of non-Archimedean nearness spaces. [ABSTRACT FROM AUTHOR]

The paper discusses mathematical concept of granular model for data and knowledge manipulation. In order to overcome the difficulties caused by extensive data representation a new model based on granular sets and granular relations is put forward. The key idea is that the notion of set may consist of basic elements grouped into bigger granules. A granular set is formed from a universe set and a semi-partition defining granules of its elements. Formal definition of granular sets and some basic algebraic operations on granular sets are introduced in the paper. Further, the concept of granular relation is also defined and some possibilities of application of granular sets and relations to knowledge representation are put forward. [ABSTRACT FROM AUTHOR]

In machine learning and knowledge discovery, rough set theory is a useful tool to be employed as a preprocessing step for dimension reduction. However, for a given system, there may be more than one reduct to be selected. Different reducts will lead to discovered knowledge, which may be concise, precise, general, understandable and practically useful in different levels. It is a crucial issue to select the most suitable features or properties of the objects in a dataset in the machine learning process. In this paper, some external information is added to information system and may be simply regarded as user preference on attributes. Consequently, it will guide the procedure of retrieving reducts, which will give birth to the reduct subject to preference order of attributes. [ABSTRACT FROM AUTHOR]

This paper concerns the development of a new direction in machine learning, called natural induction, which requires from computer-generated knowledge not only to have high predictive accuracy, but also to be in human-oriented forms, such as natural language descriptions and/or graphical representations. Such forms facilitate understanding and acceptance of the learned knowledge, and making mental models that are useful for decision making. An initial version of the AQ21-NI program for natural induction and its several novel features are briefly described. The performance of the program is illustrated by an example of deriving medical diagnostic rules from micro-array data. [ABSTRACT FROM AUTHOR]

A new fuzzy neural network is introduced in this paper which employs self-organization competition neural network to optimize the structure of the fuzzy neural network, and applies a genetic algorithm to adjust the connection weights of the fuzzy neural network so as to get the best structure and weights of the fuzzy neural network. Simulations are made when the pole becomes 2 meters and the random white noise is added in the cart-pendulum system, and control effects of the Adaptive Neural Fuzzy Illation System (ANFIS) and Genetic Algorithm Fuzzy Neural Network (GAFNN) are analyzed. Simulation results indicate that GAFNN controller has greater control performance, high convergence speed, strong robustness and better dynamic characteristics. The effectiveness of the method introduced in this paper is demonstrated by its encouraging study results. [ABSTRACT FROM AUTHOR]

To circumvent the drawbacks in nonlinear controller designing of chemical processes, an adaptive single neuron control scheme is proposed in this paper. A class of nonlinear processes is approximated by a fuzzy neural network-based model. The key of this work is, an adaptive single neuron controller, which mimics PID controller, is considered in the proposed control scheme. Applying this result and Lyapunov stability theory, a novel-updating algorithm to adjust the parameters of the single neuron controller is presented. Simulation results illustrate the effectiveness of the proposed adaptive single neuron control scheme. [ABSTRACT FROM AUTHOR]

In Multi-Agent System, the generation of agent coalition is a key topic. It mainly researches how to generate the task-oriented optimal agent coalition. This paper proposes a novel coalition generation method based on fuzzy set theory. Defines a membership function, by which the membership degree of an agent to a coalition is calculated as a value between 0 and 1, and then the task-oriented coalition can be generated by λ matrix theory. An example is presented to illustrate the validity of the method. [ABSTRACT FROM AUTHOR]

To solve the cost control problem of outsourcing procurement inherent in the complexness and difficulty for mathematical analysis, this paper established a simulation-based cost Optimization framework. Firstly, we adopted Poisson process to simulate multi-supplier outsourcing market environment, the procurement cost expectations objective function and constraints of the model was introduced. Secondly, taking decision strategy as feedback variables, procurement cost optimization model was proposed. Thirdly, several key technologies for algorithm implement were described. Finally, based on the history demand of an aircraft model factory, numerical simulation is carried out to show that this optimization model can effectively reduce the outsourcing procurement total cost and outsourcing parts inventory fluctuation. [ABSTRACT FROM AUTHOR]

In this paper, we propose the integration between Strength Pareto Evolutionary Algorithm 2 (SPEA2) with two types of coevolution concept, Competitive Coevolution (CE) and Cooperative Coevolution (CC), to solve 3 dimensional multiobjective optimization problems. The resulting algorithms are referred to as Strength Pareto Evolutionary Algorithm 2 with Competitive Coevolution (SPEA2-CE) and Strength Pareto Evolutionary Algorithm 2 with Cooperative Coevolution (SPEA2-CC). The main objective of this paper is to compare competitive against cooperative coevolution to ascertain which coevolutionary approach is preferable for multiobjective optimization. The competitive coevolution will be implemented with K-Random Opponents strategy. The performances of SPEA2-CE and SPEA2-CC for solving tri-objective problems using the DTLZ suite of test problems are presented. The results show that the cooperative approach far outperforms the competitive approach when used to augment SPEA2 for tri-objective optimization in terms of all the metrics (generational distance, spacing and coverage). [ABSTRACT FROM AUTHOR]

The paper presents a method for the identification of bilinear system parameters by using an improved Genetic Algorithm. Good results could still be obtained when the system output was influenced by Gaussian noise in the simulation. By comparing with RLS and COR through a simulation experiment to a SISO bilinear system, it is found that the method can get better result than the other two methods. Through a simulation experiment to a MIMO bilinear system, the method can get reasonably good results too. These simulations show that the method is simpler and can get better results than RLS and COR. Through a simulation study to an MIMO bilinear system, good results can still be got. In the last section, the paper describes that a hybrid GA, the combination of Genetic Algorithm and nonlinear Least Square, was developed to identify bilinear system structure and parameters simultaneously. [ABSTRACT FROM AUTHOR]

This paper describes an algorithm for generating a figure in a two-dimensional plane from a qualitative spatial representation of PLCA. In general, it is difficult to generate a figure from qualitative spatial representations, since they contain positional relationships but do not hold quantitative information such as position and size. Therefore, an algorithm is required to determine the coordinates of the objects while preserving the positional relationships. Moreover, it is more desirable that the resulting figure meets a user's requirement. PLCA is a simple symbolic representation consisting of points, lines, circuits and areas. We have already proposed one algorithm for drawing, but the resulting figures are far from a "good" one. In that algorithm, we generate the graph corresponding to a given PLCA expression, decompose it into connected subgraphs, determine the coordinates in a unit circle for each subgraph independently, and finally determine the position and size of each subgraph by locating the circles in appropriate positions. This paper aims at generating a "good" figure for a PLCA expression. We use a genetic algorithm to determine the locations and the sizes of circles in the last step of the algorithm. We have succeeded in producing a figure in which objects are drawn as large as possible, with complex parts larger than others. This problem is considered to be a type of "circle packing," and the method proposed here is applicable to the other problems in which locating objects in a non-convex polygon. [ABSTRACT FROM AUTHOR]

This paper shows how to use a top-level ontology to create robust and logically coherent domain ontology in a way that facilitates computational implementation and interoperability. It uses a domain ontology of ecosystem classification and delineation outlined informally Bailey's paper on ‘Delineation of Ecoregions' as a running example. Baily's (from an ontological perspective) rather imprecise and ambiguous definitions are made more logically rigorous and precise by (a) restating the informal definitions formally using the top-level terms whose semantics was specified rigorously in a logic-based top-level ontology and (b) by enforcing the clear distinction of types of relations as specified at the top-level and specific relations of a given type as they occur in the ecosystem domain. In this way it becomes possible to formally distinguish a number of relations which logical interrelations are important but which have been confused and been taken to be a single relation before. [ABSTRACT FROM AUTHOR]

This paper presents a method of target tracking for a robotic vision system employing reinforcement learning with feedback based on average rough coverage performance values. The application is for a line-crawling inspection robot (ALiCE II, the second revision of Automated Line Crawling Equipment) designed to automate the inspection of hydro electric transmission lines and related equipment. The problem considered in this paper is how to train the vision system to track targets of interest and acquire useful images for further analysis. To train the system, two versions of Watkins' Q-learning were implemented, the classical single-step version and a modified strain using an approximation space-based form of what we term rough feedback. The robot is briefly described along with experimental results for the two forms of the Q-learning control algorithm. The contribution of this article is an introduction to a modified version of Q-learning control with rough feedback to monitor and adjust the learning rate during target tracking. [ABSTRACT FROM AUTHOR]

This paper first presents a brief survey of the existing works on comparing and ranking any two interval numbers and then, on the basis of this, gives the inclusion measure approach to compare any two interval numbers. The monotonic inclusion measure is defined over the strict partial order set proposed by Moore and illustrate that the possibility degrees in the literature are monotonic inclusion measures defined in this paper; Then a series of monotonic inclusion measures are constructed based on t-norms. Finally, we give illustrations by using the monotonic inclusion measures and gain good results. [ABSTRACT FROM AUTHOR]

We give a description of a Petri net-based framework for modelling and analysing biochemical pathways, which unifies the qualitative, stochastic and continuous paradigms. Each perspective adds its contribution to the understanding of the system, thus the three approaches do not compete, but complement each other. We illustrate our approach by applying it to an extended model of the three stage cascade, which forms the core of the ERK signal transduction pathway. Consequently our focus is on transient behaviour analysis. We demonstrate how qualitative descriptions are abstractions over stochastic or continuous descriptions, and show that the stochastic and continuous models approximate each other. A key contribution of the paper consists in a precise definition of biochemically interpreted stochastic Petri nets. Although our framework is based on Petri nets, it can be applied more widely to other formalisms which are used to model and analyse biochemical networks. [ABSTRACT FROM AUTHOR]

The paper proposes the use of the multilayer perceptron model to the problem of detecting attack patterns in computer networks. The multilayer perceptron is trained and assessed on patterns extracted from the files of the Third International Knowledge Discovery and Data Mining Tools Competition. It is required to classify novel normal patterns and novel categories of attack patterns. The results are presented and evaluated in the paper. [ABSTRACT FROM AUTHOR]

A novel hierarchical hybrid neural model to the problem of long-term electrical load forecasting is proposed in this paper. The neural model is made up of two self-organizing map nets — one on top of the other —, and a single-layer perceptron. It has application into domains which require time series analysis. The model is compared to a multilayer perceptron. Both the hierarchical and the multilayer perceptron models are endowed with time windows in their input layers. They are trained and assessed on load data extracted from a North-American electric utility. The models are required to predict once every week the electric peak-load and mean-load during the next two years. The results are presented and evaluated in the paper. [ABSTRACT FROM AUTHOR]

This paper presents a hardware implementation of a Time Multiplexing Architecture (TMA) that can interconnect arrays of neurons in an Artificial Neural Network (ANN) using a single metal wire. The approach exploits the relative slow operational speed of the biological system by using fast digital hardware to sequentially sample neurons in a layer and transmit the associated spikes to neurons in other layers. The motivation for this work is to develop minimal area inter-neuron communication hardware. An estimate of the density of on-chip neurons afforded by this approach is presented. The paper verifies the operation of the TMA and investigates pulse transmission errors as a function of the sampling rate. Simulations using the Xilinx System Generator (XSG) package demonstrate that the effect of these errors on the performance of an SNN, pre-trained to solve the XOR problem, is negligible if the sampling frequency is sufficiently high. [ABSTRACT FROM AUTHOR]

The paper deals with the physical implementation of biologically plausible spiking neural network models onto a hardware architecture with bio-inspired capabilities. After presenting the model, the work will illustrate the major steps taken in order to provide a compact and efficient digital hardware implementation of the model. Special emphasis will be given to the scalability features of the architecture, that will permit the implementation of large-scale networks. The paper will conclude with details about the physical mapping of the model, as well as with experimental results obtained when applying dynamic input stimuli to the implemented network. [ABSTRACT FROM AUTHOR]

Compact representation of knowledge having strong internal interactions has become possible with the developments in neurocomputing and neural information processing. The field of neural networks has offered various solutions for complex problems, however, the problems associated with the learning performance has constituted a major drawback in terms of the realization performance and computational requirements. This paper discusses the use of variable structure systems theory in learning process. The objective is to incorporate the robustness of the approach into the training dynamics, and to ensure the stability in the adjustable parameter space. The results discussed demonstrate the fulfillment of the design specifications and display how the strength of a robust control scheme could be an integral part of a learning system. This paper discusses how Gaussian radial basis function neural networks could be utilized to drive a mechatronic system's behavior into a predefined sliding regime, and it is seen that the results are promising. Keywords: Gaussian Radial Basis Function Networks, Sliding Mode Control. [ABSTRACT FROM AUTHOR]

Neural networks represent a class of functions for the efficient identification and forecasting of dynamical systems. It has been shown that feedforward networks are able to approximate any (Borel-)measurable function on a compact domain [1,2,3]. Recurrent neural networks (RNNs) have been developed for a better understanding and analysis of open dynamical systems. Compared to feedforward networks they have several advantages which have been discussed extensively in several papers and books, e.g. [4]. Still the question often arises if RNNs are able to map every open dynamical system, which would be desirable for a broad spectrum of applications. In this paper we give a proof for the universal approximation ability of RNNs in state space model form. The proof is based on the work of Hornik, Stinchcombe, and White about feedforward neural networks [1]. [ABSTRACT FROM AUTHOR]

Quantifying the quality of a video or audio transmission over the Internet is usually a hard task, as based on the statistical processing of the evaluations made by a panel of humans (the corresponding and standardized area is called subjective testing). In this paper we describe a methodology called Pseudo-Subjective Quality Assessment (PSQA), based on Random Neural Networks, which is able to perform this task automatically, accurately and efficiently. RNN had been chosen here because of their good performances over other possibilities; this is discussed in the paper. Some new insights on PSQA's use and performance are also given. In particular we discuss new results concerning PSQA-based dynamic quality control, and conversational quality assessment. [ABSTRACT FROM AUTHOR]

This paper presents the dense areas based algorithm for generating fuzzy rules for classification WWW documents. Description document clusters in the form of fuzzy rules (FR) make possible the presentation of information in the form fuzzy granules. Moreover, each cluster might be described by several fuzzy rules. These fuzzy rules can be used as the knowledge base for searching new information from WWW resources with regard to specific topics and users' requirements. [ABSTRACT FROM AUTHOR]

This paper demonstrates that AI methods - in particular, linguistic mechanisms of semantic meaning reasoning can be applied to the development of intelligent IT systems. They also facilitate an in-depth analysis of the meaning presented in DDS information systems. This paper also presents the IT mechanisms of object meaning description on selected examples of spinal cord image analysis. The procedures for such semantic reasoning are based on the model of cognitive resonance. These have been applied to the task of interpreting the meaning of selected diagnostic images from the central nervous system as an intelligent analysis module in IT systems. The application presented in this paper is of a research character and it serves the preparation of efficient lesion detection methods applied to a dataset originating from magnetic and resonance examinations of the spinal cord structures. [ABSTRACT FROM AUTHOR]

In this paper, we propose multi-class classifier and knowledge based face detection. Eye region and face location is used illuminant based Bayesian detector. We propose the efficient face and eye detection system using varying illuminant context modeling and multi-classifier. The face detection system architecture use cascade method by illuminant face model. Also, we detect eye region after face detection. Proposed eye detection frame is multiple illuminant Bayesian classifiers. Because face images have varying illuminant and this is vary difficult problem in face detection. Therefore, we made in context model using face illuminant. The multiple classifiers consist of face illuminant information. Multiple Bayesian classifiers are employed for selection of face and eye detection windows on illuminant face group. Finally, face and eye regions of the detected candidates are selected by context awareness. [ABSTRACT FROM AUTHOR]

In this paper, we present a novel approach to learning from visual information given in a form of raster images. The proposed learning method uses genetic programming to synthesize an image processing procedure that performs the desired vision task. The evolutionary algorithm maintains a population of individuals, initially populated with random solutions to the problem. Each individual encodes a directed acyclic graph, with graph nodes corresponding to elementary image processing operations (like image arithmetic, convolution filtering, morphological operations, etc.), and graph edges representing the data flow. Each graph contains a single input node to feed in the input image and an output node that yields the final processing result. This genetic learning process is driven by a fitness function that rewards individuals for producing output that conforms the task-specific objectives. This evaluation is carried out with respect to the training set of images. Thanks to such formulation, the fitness function is the only application-dependent component of our approach, which is thus applicable to a wide range of vision tasks (image enhancement, object detection, object tracking, etc.). The paper presents the approach in detail and describes the computational experiment concerning the task of object tracking in a real-world video sequence. [ABSTRACT FROM AUTHOR]

To take into account different character of distinct segments of non-stationary financial time series the multi-agent system based forecasting algorithm is suggested. The primary goal of present paper is to introduce methodological findings that could help to reduce one step ahead forecasting error. In contrast to previous investigation [6], instead of single prediction rule we use a system of several adaptive forecasting agents. The agents evolve, compete among themselves. Final decision is made by a collective of the most successive agents and present time moment. New multi-agent forecasting system allows utilizing shorter training sequences and results in more accurate forecasts than employing single prediction algorithm. Keywords: Classification, Forecasting, Sliding window, Training, Dimensionality. [ABSTRACT FROM AUTHOR]

This paper describes a technique for statically stable gait synthesis for a quadruped robot using a simple Feed Forward Neural Networks (FFNN). A common approach for gait synthesis based on neural networks, is to use an implementation with Continuous Time Recurrent Neural Network (CTRNN) of arbitrary complex architecture as pattern generator for rhythmic limb motion. The preferred training method is implemented using genetic algorithms (GAs). However, to achieve the desired trajectory becomes an obstacle during the training process. This paper presents a much more simpler process converting a statically stable gait into actuator's space via inverse kinematics; the training of the network is done with those references. By doing so, the training problem becomes a spatio-temporal machine learning problem. It is described a solution for trajectory generation combining a simple oscillator model with a Multilayer Feedforward Neural Network (MFNN) to generate the desired trajectory. [ABSTRACT FROM AUTHOR]

In this paper, a concept of directional mutations for phenotypic evolutionary algorithms is presented. The proposed approach allows, in a very convenient way, to adapt the probability measure underlying the mutation operator during evolutionary process. Moreover, the paper provides some guidance, along with suitable theorems, which makes it possible to get a deeper understanding of the ineffectiveness of isotropic mutations for large-scale problems. [ABSTRACT FROM AUTHOR]

Growing expectations for low priced software-intensive applications tailored to the individual customers needs require the usage of techniques aimed at the individual development of application systems, which could be attained at the costs of standard software and easily adapted. In the paper the concepts associated with assembling the applications by means of software factories in conjunction with the fuzzy description of the variable product parts are introduced. The approach is demonstrated in the example of a part of an order process. [ABSTRACT FROM AUTHOR]

Modern and effective methods of knowledge extraction from databases and information systems are required to provide rather linguistic than numerical information. The so-called linguistic summaries of databases by Yager [1], exemplified by Many children like sweet ice cream, and further improvements by George and Srikanth [2] and by Kacprzyk and Yager [3], are discussed in this paper. The use of type-2 fuzzy sets is an original contribution to the domain, since only ordinary fuzzy sets have been originally employed. Elements of type-2 semantics are shown to handle effectively pieces of imprecise information (e.g. fuzzy sets) stored in databases. An application on sample data is provided. [ABSTRACT FROM AUTHOR]

We propose a method of multiple context fusion based robust face detection scheme, multiple cascade and finally decision using correlation table. It takes advantage of multiple cascade face detector fusion by context. We propose the filtering classifier method for illumination face image. And then we constructed cascade classifier from applied different filtering method. The multiple cascade detectors made from six single context detectors. Six contexts are divided k-means algorithm, and classify illuminant. In this paper, we proposed the classifier fusion method by using correlation between face images. 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. We achieved very encouraging experimental results having varying illuminant. [ABSTRACT FROM AUTHOR]

Retrieving pertinent parts of a meeting or a conversation recording can help for automatic summarization or indexing of the document. In this paper, we deal with an original task, almost never presented in the literature, which consists in automatically extracting questions utterances from a recording. In a first step, we have tried to develop and evaluate a question extraction system which uses only acoustic parameters and does not need any textual information from a speech-to-text automatic recognition system (called ASR system for Automatic Speech Recognition in the speech processing domain) output. The parameters used are extracted from the intonation curve of the speech utterance and the classifier is a decision tree. Our first experiments on French meeting recordings lead to approximately 75% classification rate. An experiment in order to find the best set of acoustic parameters for this task is also presented in this paper. Finally, data analysis and experiments on another French dialog database show the need of using other cues like the lexical information from an ASR output, in order to improve question detection performance on spontaneous speech. [ABSTRACT FROM AUTHOR]