Knowledge-Based Intelligent Information and Engineering Systems

We report some of our results of a particular blind source separation technique applied to spectral unmixing of remote-sensed hyperspectral images. Different nongaussianity measures are introduced in the learning procedure, and the results are compared to assess their relative efficiencies, with respect to both the output signal-to-interference ratio and the overall computational complexity. This study has been conducted on both simulated and real data sets, and the first results show that skewness is a powerful and unexpensive tool to extract the typical sources that characterize remote-sensed images.

A second-order statistical technique (

FD-CCA

) for semi-blind source separation from multiple-sensor data is presented. It works in the Fourier domain and allows us to both learn the unknown mixing operator and estimate the source cross-spectra before applying the proper source separation step. If applied to small sky patches, our algorithm can be used to extract diffuse astrophysical sources from the mixed maps obtained by radioastronomical surveys, even though their resolution depends on the measurement channel. Unlike the independent component analysis approach,

FD-CCA

does not need mutual independence between sources, but exploits their spatial autocorrelations. We describe our algorithm, derived from a previous pixel-domain strategy, and present some results from simulated data.

This paper extends the use of stochastic watershed, recently introduced by Angulo and Jeulin [1], to unsupervised segmentation of multispectral images. Several probability density functions (pdf), derived from Monte Carlo simulations (

M

realizations of

N

random markers), are used as a gradient for segmentation: a weighted marginal pdf a vectorial pdf and a probabilistic gradient. These gradient-like functions are then segmented by a volume-based watershed algorithm to define the

R

largest regions. The various gradients are computed in multispectral image space and in factor image space, which gives the best segmentation. Results are presented on PLEIADES satellite simulated images.

In this paper we consider the problem of separating source images from linear mixtures with unknown coefficients, in presence of noise and blur. In particular, we consider as a special case the problem of estimating the Cosmic Microwave Background from galactic and extra–galactic emissions. Like many visual inverse problems, this problem results to be ill–posed in Hadamard sense. To solve the non–blind version of the problem a classical edge–preserving regularization technique can be used. Thus, the solution is defined as the argument of the minimum of an energy function. In order to solve the blind inverse problem, in this paper a new function, called target function, is introduced. Such a function can consider constraints as the degree of Gaussianity and correlation of the results. The experimental results, considering the cosmic mixtures, have given accurate estimations.

We present an unsupervised method for fully automatic detection of regions of interest containing fibroglandular tissue in digital screening mammography. The unsupervised segmenter is based on a combination of several unsupervised segmentation results, each in different resolution, using the sum rule. The mammogram tissue textures are locally represented by four causal monospectral random field models recursively evaluated for each pixel. The single-resolution segmentation part of the algorithm is based on the underlying Gaussian mixture model and starts with an over segmented initial estimation which is adaptively modified until the optimal number of homogeneous mammogram segments is reached. The performance of the presented method is extensively tested on the Digital Database for Screening Mammography (DDSM) from the University of South Florida as well as on the Prague Texture Segmentation Benchmark using the commonest segmentation criteria and where it compares favourably with several alternative texture segmentation methods.

Localization is one of the important topics in robotics and it is essential to execute a mission. Most problems in the class of localization are due to uncertainties in the modeling and sensors. Therefore, various filters are developed to estimate the states in noisy information. Recently, particle filter is issued widely because it can be applied to a nonlinear model and a non-Gaussian noise. In this paper a fuzzy adaptive particle filter is proposed, whose basic idea is to generate samples at the high-likelihood using a fuzzy logic approach. The method brings out the improvement of an accuracy of estimation. In addition, this paper presents the localization method for a mobile robot with ultrasonic beacon systems. For comparison purposes, we test a conventional particle filter method and our proposed method. Experimental results show that the proposed method has better localization performance.

In this paper, we introduce a new fuzzy model called fuzzy combined polynomial neural networks, which are based on the representative fuzzy model named polynomial fuzzy model. In the design procedure of the proposed fuzzy model, the coefficients on consequent parts are estimated by using not general least square estimation algorithm that is a sort of global learning algorithm but weighted least square estimation algorithm, a sort of local learning algorithm. We are able to adopt various type of structures as the consequent part of fuzzy model when using a local learning algorithm. Among various structures, we select Polynomial Neural Networks which have nonlinear characteristic and the final result of which is a complex mathematical polynomial. The approximation ability of the proposed model can be improved using Polynomial Neural Networks as the consequent part.

The human-machine interface is an essential part of intelligent robotic system. Through the human-machine interface, human being can interact with the robot. Especially, in tele-robotics environment, the human-machine interface can be developed with remarkable extended functionality. In this paper, we propose a human-machine interface with augmented reality for the network based mobile robot. Generally, we can take some meaningful information from human’s motion such as movement of head or fingers. So, it is very useful to take these motions as input for systems. We synchronize head motion of human being and the camera motion of the mobile robot using visual information. So user of the mobile robot can monitor environment of the mobile robot as eyesight of mobile robot. Then we use gesture recognition for control the mobile robot. In the implemented framework, the user can monitor what happens in environment as eyesight of mobile robot and control the mobile robot easily and intuitively by using gesture.

This paper presents an efficient method of obstacle recognition system and HRI (human robot interaction) system specialized for biped walking robot. This method transmits the information regarding obstacle conditions to a biped walking robot. In the present paper, we describe a cascade of boosted classifier using adaboost algorithm as a obstacle region extracting module from input images. Besides, PCA is applied as a feature extracting module from the obstacle region and a hierarchical support vector machine is applied as an obstacle recognizing module. The data from vision system is combined with information from other sensors and the walking assist commands transmit to the biped walking robot. From the results of experiments, the proposed method can be applied to biped walking robot effectively.

This paper presents an original feature selection method for Emotion Recognition which includes many original elements. Feature selection has some merit regarding pattern recognition performance. Thus, we developed a method called an ’Interactive Feature Selection(IFS)’ and ’GA Feature Selection( GAFS)’. Afterwards, the results (selected features) of the IFS and GAFS were applied to an emotion recognition system (ERS), which was also implemented in this research. Especially, our interactive feature selection method was based on a Reinforcement Learning Algorithm since it required responses from human users. By performing the IFS, we were able to obtain three top features and apply them to the ERS. We compared those results from a random selection and Sequential Forward Selection (SFS) and Genetic Algorithm Feature Selection (GAFS).

Learning behaviors of the hierarchical structure learning automata (HSLA) with the three representative algorithms under the nonstationary multiteacher environments are considered. Several computer simulations confirm the effectiveness of the newly developed relative reward strength algorithm (NRRSA).

This paper presents an improved statistical learning algorithm for feature generation in pattern recognition and signal processing. It is approximately derived from geometrical interpretation of the Fisher linear discriminant analysis (FLDA). The principal component analysis (PCA) is popular for data compression and feature extraction. Furthermore, iterative learning algorithms for obtaining eigenvectors in PCA have been presented in such fields. Their effectiveness has been demonstrated in many applications. However, recently FLDA has been often used in many fields, especially face image recognition. The drawback of FLDA is a long computational time based on a large-sized covariance matrix and the issue that the within-class covariance matrix is usually singular. Generally, in FLDA, the inverse matrix of the within-class covariance matrix cannot be obtained, since data dimension is generally higher than the number of data and then it includes many zero eigenvalues. In order to overcome this difficulty, a new iterative feature generation method, the simple-FLDA was proposed by authors. In this paper, further improvement is introduced into the simple-FLDA and its effectiveness is demonstrated for preliminary personal identification problem.

In this paper, we discuss a robot vision in order to perceive humans and the environment around a mobile robot. We developed a tele-operated mobile robot with a pan-tilt mechanism composed of a camera and a laser range finder (LRF). The output from the camera is color information, and the output of LRF is distance information to objects from the robot. In this paper, we propose a method of sensor fusion to extract a human from the measured data by integrating these outputs based on the concept of synthesis. Finally, we show experimental results of the proposed method.

Learning Automata (LA) were recently shown to be valuable tools for designing Multi-Agent Reinforcement Learning algorithms. One of the principal contributions of LA theory is that a set of decentralized, independent learning automata is able to control a finite Markov Chain with unknown transition probabilities and rewards. We extend this result to the framework of Multi-Agent MDP’s, a straightforward extension of single-agent MDP’s to distributed cooperative multi-agent decision problems. Furthermore, we combine this result with the application of parametrized learning automata yielding global optimal convergence results.

Collective behavior such as bird flocking, land animal herding, and fish schooling is well known in nature. Many observations have shown that there are no leaders to control the behavior of a group. Several models have been proposed for describing the grouping behavior, which we regard as a distinctive example of aggregate motions. In these models, a fixed rule is provided for each of the individuals a priori for their interactions in a reductive and rigid manner. In contrast, we propose a new framework for the self-organized grouping of agents by reinforcement learning. It is important to introduce a learning scheme for causing collective behavior in artificial autonomous distributed systems. The behavior of agents is demonstrated and evaluated through computer simulations and it is shown that their grouping and anti-predator behavior emerges as a result of learning.

In this paper, systematic three-dimensional classification is presented for sets of interneuron slice images of silkworm moths, using self-organizing maps. Fractal dimension values are calculated for target sets to quantify denseness of their branching structures, and are employed as element values in training data for constructing a map. The other element values are calculated from the sets to which labeling and erosion are applied, and they quantifies whether the sets include thick main dendrites. The classification result is obtained as clusters with units in the map. The proposed classification employing only two elements in training data achieves as high accuracy as the manual classification made by neuroscientists.

We propose a modified SVM algorithm for the classification of data augmented with explicit quality quantification for each example in the training set. As the extension to nonlinear decision functions through the use of kernels brings to a non-convex optimization problem, we develop an approximate solution. Finally, the proposed approach is applied to a set of benchmarks and contrasted with analogous methodologies in the literature.

Sensor networks have become an important source of data with numerous applications in monitoring various real-life phenomena as well as industrial applications and traffic control. However, sensor data are subject to several sources of errors as the data captured from the physical world through these sensor devices tend to be incomplete, noisy, and unreliable, thus yielding imprecise or even incorrect and misleading answers which can be very significative if they result in immediate critical decisions or activation of actuators. Traditional data cleaning techniques cannot be applied in this context as they do not take into account the strong spatial and temporal correlations typically present in sensor data, so machine learning techniques could greatly be of aid. In this paper, we propose a neuro-fuzzy regression approach to clean sensor network data: the well known ANFIS model is employed for reducing the uncertainty associated with the data thus obtaining a more accurate estimate of sensor readings. The obtained cleaning results show good ANFIS performance compared to other common used model such as kernel methods, and we demonstrate its effectiveness if the cleaning model has to be implemented at sensor level rather than at base-station level.

A new approach of input uncertainty classification is proposed in this paper. This approach develops a new technique which extends the support vector classification (SVC) by incorporating input uncertainties. Kernel functions can be used to generalize this proposed technique to non-linear models and the resulting optimization problem is a second order cone program with a unique solution. Results are shown to demonstrate how the technique is more robust when uncertainty information is available.

We present a novel approach for Fuzzy-Input Fuzzy-Output classification. One-Against-All Support Vector Machines are adapted to deal with the fuzzy memberships encoded in fuzzy labels, and to also give fuzzy classification answers. The mathematical background for the modifications is given. In a benchmark application, the recognition of emotions in human speech, the accuracy of our F

2

-SVM approach is clearly superior to that of fuzzy MLP and fuzzy K-NN architectures.

Deterministic search algorithm such as greedy search is apt to get into local maxima, and learning Bayesian networks (BNs) by stochastic search strategy attracts the attention of many researchers. In this paper we propose a BN learning approach, E-MDL, based on stochastic search, which evolves BN structures with an evolutionary algorithm and can not only avoid getting into local maxima, but learn BNs with hidden variables. When there exists incomplete data, E-MDL estimates the probability distributions over the local structures in BNs from incomplete data, then evaluates BN structures by a variant of MDL score. The experimental results on Alarm, Asia and an examplar network verify the validation of E-MDL algorithm.

A proper theoretical framework, called

reliable learning

, for the analysis of consistency of learning techniques incorporating prior knowledge for the solution of pattern recognition problems is introduced by properly extending standard concepts of Statistical Learning Theory.

In particular, two different situations are considered: in the first one a reliable region is determined where the correct classification is known; in the second case the prior knowledge regards the correct classification of some points in the training set. In both situations sufficient conditions for ensuring the consistency of the Empirical Risk Minimization (ERM) criterion is established and an explicit bound for the generalization error is derived.

We devise an SVM for partitioning a sample space affected by random binary labels. In the hypothesis that a smooth, possibly symmetric, conditional label distribution graduates the passage from the all 0-label domain to the all 1-label domain and under other regularity conditions, the algorithm supplies an estimate of the above probabilities. Within the Algorithmic Inference framework, the randomness of the labels maintains the main features of the binary classification problem, yet adding a further dimension to the search space. Namely the new dimension of each point in the original space hosts the uniform seeds accounting for the randomness of the labels, so that the problem becomes that of separating the points in the augmented space. We solve it with a new kind of bootstrap technique. As for error bounds of the proposed algorithm, we obtain confidence intervals that are up to an order narrower than those supplied in the literature. This benefit comes from the fact that: (i) we devise a special algorithm to take into account the random profile of the labels; (ii) we know the number of support vectors really employed, as an ancillary output of the learning procedure; and (iii) we can appreciate confidence intervals of misclassifying probability exactly in function of the cardinality of these vectors. We numerically check these results by measuring the coverage of the confidence intervals.

This paper proposes a neural-based method to map facial expressions between human and robot. We applied the method to a sensitivity communication robot, Ifbot, which has been developed by our industry-university joint research project. The method enables the robot to imitate an arbitrary human facial expression. The paper describes the feature extraction from face image, and proposes neural network based parameter matching between human facial expression and Ifbot’s facial expression mechanism. This paper also reports the evaluation of the facial expression transmission performance of Ifbot with the proposed system. The evaluation shows the effectiveness of emphasizing emotional expressions and the possibility of using Ifbot as an agent for distance communication.

Automatic topic extraction from a large number of documents is useful to capture an entire picture of the documents or to classify the documents. Here, it is an important issue to evaluate how much the extracted topics, which are set of documents, are interpretable for human. As the objective is vector representable topic extractions, e.g., Latent Semantic Analysis, we tried to formulate the interpretable likelihood of the extracted topic using the manually derived topics. We evaluated this likelihood of topics on English news articles using

LSA

,

PCA

and Spherical

k

-means for topic extraction. The results show that this likelihood can be applied as a filter to select meaningful topics.

This paper addresses learning of a reasonably accurate evaluation function of Shogi (Japanese Chess) positions through learning from records of games. Accurate evaluation of a Shogi position is indispensable for a computer Shogi program. A Shogi position is projected into several semantic features characterizing the position. Using such features as input, we employ reinforcement learning with a multi-layer perceptron as a nonlinear function approximator. We prepare two completely different sets of games: games played by computer Shogi programs and games played by professional Shogi players. Then we built two evaluation functions by separate learning based on two different sets of games, and compared the results to find several interesting tendencies.

We present a method of nominally piecewise multiple regression using a four-layer perceptron to fit multivariate data containing numerical and nominal variables. In our method, each linear regression function is accompanied with the corresponding nominal condition stating a subspace where the function is applied. Our method selects the optimal numbers of hidden units and rules very fast based on the Bayesian Information Criterion (BIC). The proposed method worked well in our experiments using an artificial and two real data sets.

Similarity search, finding objects similar to a given query object, is an important operation in multimedia databases, and has many applications in a wider variety of fields. As one approach to efficient similarity search, we focus on utilizing a set of pivots for reducing the number of similarity calculations between a query and each object in a database. In this paper, unlike conventional methods based on combinatorial optimization, we propose a new method for learning a set of pivots from existing data objects, in virtue of iterative numerical nonlinear optimization. In our experiments using one synthetic and two real data sets, we show that the proposed method significantly reduced the average number of similarity calculations, compared with some representative conventional methods.

We address the problem of predicting link attachments to complex networks. As one approach to this problem, we focus on combining network growth (or information propagation) models with machine learning techniques. In this paper, we present a method for predicting link conversions based on the estimated probability of information propagation on each link. In our experiments using a real blogroll network, we show that the proposed method substantially improved the predictive performance based on the F-measure, in comparison to other methods using some conventional criteria.

The main purpose of a Non Destructive Evaluation technique is to provide information about the presence/absence, Within this framework, it is very important to automatically detect and characterize defect minimizing the indecision about measurements. This paper just treats an inverse electrostatic problem, with the aim of detecting and characterizing semi-spherical defects (i.e. superficial defects) on metallic plates. Its originality consists on the proposed electromagnetic way exploited to a non destructive inspection of specimens as well as on the use of a Support Vector Regression Machine based approach in order to characterize the detected defect. The experimental results show the validity of the proposed processing.

In this paper, we propose a compensation algorithm to reduce the nonlinearity error which is occurred in a heterodyne laser interferometer as a nano-meter scale measurement apparatus. In heterodyne laser interferometer, frequency-mixing is the main factor of nonlinearity error. Using an RLS algorithm, the nonlinearity compensation parameters are found to be used through geometric projection. With the roughly modified intensity signals from LIA, the back-propagation neural network algorithm minimizes the objective function to track the reference signal for learning period. Through some experiments, it is verified that the proposed algorithm can reduce nonlinear factors and improve the measurement accuracy of laser interferometer.

The study described in this paper, analyzed the urban and suburban air pollution principal causes and identified the best subset of features (meteorological data and air pollutants concentrations) for each air pollutant in order to predict its medium-term concentration (in particular for the PM

10

). An information theoretic approach to feature selection has been applied in order to determine the best subset of features by means of a proper backward selection algorithm. The final aim of the research is the implementation of a prognostic tool able to reduce the risk for the air pollutants concentrations to be above the alarm thresholds fixed by the law. The implementation of this tool will be carried out using machine learning methods based on some of the most wide-spread statistical data driven techniques (Artificial Neural Networks, ANN, and Support Vector Machines, SVM).

This paper presents a handwritten Greek character recognizer. The recognizer is composed of two modules: the first one is a feature extractor, the second one, the classifier, is performed by means of

Learning Vector Quantization

. The recognizer, tested on a database of more than 28000 handwritten Greek characters, has shown satisfactory performances.

This paper reports on the unsupervised analysis of seismic signals recorded in Italy, respectively on the Vesuvius volcano, located in Naples, and on the Stromboli volcano, located North of Eastern Sicily. The Vesuvius dataset is composed of earthquakes and false events like thunders, man-made quarry and undersea explosions. The Stromboli dataset consists of explosion-quakes, landslides and volcanic microtremor signals. The aim of this paper is to apply on these datasets three projection methods, the linear Principal Component Analysis (PCA), the Self-Organizing Map (SOM), and the Curvilinear Component Analysis (CCA), in order to compare their performance. Since these algorithms are well known to be able to exploit structures and organize data providing a clear framework for understanding and interpreting their relationships, this work examines the category of structural information that they can provide on our specific sets. Moreover, the paper suggests a breakthrough in the application area of the SOM, used here for clustering different seismic signals. The results show that, among the three above techniques, SOM better visualizes the complex set of high-dimensional data discovering their intrinsic structure and eventually appropriately clustering the different signal typologies under examination, discriminating the explosion-quakes from the landslides and microtremor recorded at the Stromboli volcano, and the earthquakes from natural (thunders) and artificial (quarry blasts and undersea explosions) events recorded at the Vesuvius volcano.

For conventional ships, the mono-variable autopilot controls the heading of the ship in the presence of disturbances. During the heading control, there are many moments of time when the rudder command to control the yaw angle has a negative influence on roll oscillations. The prediction of the wave influence on the roll motion can be used to implement an intelligent heading control system, which is added to the mono-variable autopilot, generating only rudder commands with damping or non-increasing effects over roll movements. In this paper, aspects of roll angle and roll rate prediction using feed-forward neural networks are discussed. A neural network predictor of the roll rate, based on measured values of the roll angle, is proposed. The neural architecture is analyzed using different training data sets and noise conditions. The predictor has on-line adaptive characteristics and is working well even if both training and testing sets are affected by measurement noise.

This study is a part of an Italian national project named STIPER, whose aim is the design and development of devices to help blind people in daily activities. This device acquires images of printed text, recognizes it through a set of Artificial Neural Networks and drives a Braille matrix or a speech synthesis engine in order to allow the user read text in real time by means of a common PDA device.

A key problem in time series prediction using autoregressive models is to fix the

model order

, namely the number of past samples required to model the time series adequately. The estimation of the model order using cross-validation is a long process. In this paper we explore faster alternative to cross-validation, based on nonlinear dynamics methods, namely Grassberger-Procaccia, Kégl and False Nearest Neighbors algorithms. Once the model order is obtained, it is used to carry out the prediction, performed by a SVM. Experiments on three real data time series show that nonlinear dynamics methods have performances very close to the cross-validation ones.

This paper presents a new model inspired from DNA bio-chemical procedures to rearrange a robot in a flexible manufacturing system to assemble different products. In a bio-chemical wet laboratory, DNA process can provide powerful massive parallelism with energy efficiency in overall search of all possible solutions. These characteristics are useful especially when dealing with a complex calculation e.g. an NP hard problem such as to a scheduling task in production line. We present a detailed algorithm and illustrate how the essential procedure for scheduling problem is realized through a DNA computing mechanism. Some experiments were conducted to show the performance of the proposed approach.

We propose an identification method of fuzzy measures by diamond pairwise comparisons. Right and left side of the diamond means ordinal weights’ comparison and up and down means interaction degrees. From the comparisons, we estimate a hierarchy diagram with interaction degrees and weights of evaluation items and the fuzzy measure of the diagram. In this paper, we compare the two interpretations of the diamond, AHP scales and Grabish’s graphical interpretation.

In this study, we apply the self-adaptive evolutionary negative selection approach for home abnormal events detection. The negative selection algorithm, also termed the exhaustive detector generating algorithm, is for various anomaly detection problems, and the concept originates from artificial immune system. Regarding the home abnormal control rules as the detector, we apply fuzzy genetic algorithm for self-adaptive information appliances control system, once the environment factors change. The proposed approach can be adaptive and incremental for the home environment factor changes. Via implementing the proposed approach on the abnormal temperature detection, we can make the information appliance control system more secure, adaptive and customized.

Using the nonparametric investment weighted kernel density approach, the study presents an alternative measure on Chinese urbanization distribution, which takes a new perspective on how investment promoting urbanization. It proposes projects to promote urbanization as an application. The projects illustrate that the overall urbanization can be promoted by 2 percent if make a 10 percent rise of investment put into the cities with urbanization levels higher than the average level; the same effect can be gained without any additional investment by changing 50 percent of the original added value of investment in the cities.

The objective of this paper is to find the inherent factors in feelings while wearing a brassiere and to consider the possibility to simulate or to predict them based on naked breast shape. For this purpose, first we conducted an experiment in which 37 Japanese women were 3D measured in naked and brassiere-wearing postures and responded to a 5-grade SD evaluation questionnaire with 13 items about feelings in a brassiere. They tried three types of brassieres; a full-cup, a 3/4-cup and a half-cup brassiere. The factor analysis revealed three factors in wearing feelings; discomfort with a wire under the bust, pressure sensation and fittedness sensation. Then the factor scores and evaluation scores were examined whether they could be predicted using breast shape data extracted from the 3D body shape models of the subjects without a brassiere and from their simulated brassiere-wearing body shape. The genetic algorithm was adopted to search the best predicting multi-regression function of control point coordinate values of the model. Searched functions were assessed according to their coefficients of correlation between the function values and those to be predicted. The final results showed that feelings about two evaluation items related to fittedness sensation could be simulated with a correlation coefficient of 0.6 to 0.8.

Grid computing architecture was defined to be a complete physical layer. The data transfer in network must be in secure. In this study, we propose the encryption algorithm in every grid node to keep information in security. User sends user-id and password to supervisor node, then supervisor node sends to authority node to certificate. Supervisor receives message from authority node and return to user node. When these algorithms install in all grid nodes, we can keep authentication process be secure in all system.

This paper focuses on the customer behaviors of Taiwan non-store channels which are "Television Commerce", "Direct Mail and Catalog Marketing", and "Internet Commerce/e-shopping". A survey is conducted to compare consumers’ characteristics, motivations and differentiated between the stages of the buying process and behavior among those three channels. Finally, the results and suggestions are stated as an important reference for non-store retailers while developing the marketing strategies.

Ever since scientists discovered that conventional silicon-based computers have an upper limit in terms of speed, they have been searching for alternative media with which to solve computational problems. That search has led them, among other places, to DNA. In this paper, we briefly review the recent developments in bio-soft computing towards the management engineering problems. The challenge and future potential of DNA computing is also addressed.

The aim of this study is to construct a multi-camera tracking system which recognizes human motions beyond several video scenes. Generally, to recognize a single human motion is easier than to link several human motions. This is because several human motions move different directions, while a single human motion moves one direction in a certain period. Therefore, we need a system that is reliable to track of human motions in every frame. In this paper, a new method is proposed for a human tracking system in order to link human movements in different directions using foot step direction method. The detail outcome and result are discussed using experiments of the method in this paper.

Facility location decisions are a critical element in strategic planning for multinational enterprises (MNEs). In this study, we propose a group fuzzy assessment method to tackle the facility location decisions for MNEs based on investment environment factors. Via the proposed group assessment model, our result will be more objective and unbiased since it is generated by a group of evaluators.

Human feelings are very complicated and it is not easy to evaluate them. Methods of paired comparison might be useful to evaluate human feelings because it is generally said that human feelings are nonlinear and methods of of paired comparison are considered to be able to deal with nonlinear problems. In AHP(Analytic Hierarchy Process) which uses paired comparison consistency of subjects” answers is judged using consistency index referred to as C.I. in short and threshold of C.I. on consistency of subjects’ answers is generally 0.1 or 0.15 by empirical judgment. This threshold has no theoretical or experimental ground. Here it is studied to determine proper threshold of C.I. comparing the coefficient of consistency which is the measurement on consistency of subjets’ answers in paired comparison.

We conducted an experiment to measure the dependence on a collision alarm system by using a driving simulator. Each participant must pass through intersections where some crossing cars may appear. A collision alarm system sounds an alarm for a crossing car. We measured both changes in driver’s dependence on the system when the alarm system provides correct and incorrect information.

This paper aims at using a numerical simulation method to discuss the influence of the differences in designs of orifice geometries on coherent of the waterjets while cutting accurately with the high velocity waterjets system. In the study process, we set up the numerical models of standard waterjets including the water container of the pressure, orifice housing, and focusing tube at first. Then, infer the governing equations system that institute of computing needs (Two-dimensional axisymmetry system). Finally, according to the different conditions to simulate the influences on velocity distributions of waterjets when the designs of the orifice produce the centre line offset and lead edge cutting distance asymmetrically. The above-mentioned results of study are not only contributing to understanding the high velocity waterjets applying to practical accurate cuttings, but also it can be references on shipbuilding and relevant industries to develop and to design on watejets orifices.

This paper addresses the issue of learning directly from the observed data in Blind Source Separation (BSS), a particular inverse problem. This problem is very likely to occur when we are dealing with two or more independent electromagnetic sources. A powerful approach to BSS is Independent Component Analysis (ICA). This approach is much more powerful if no apriori assumption about data distribution is made: this is possible transferring as much information as possible to the learning machine defining a cost function based on an information theoretic criterion. In particular, Renyi’s definition of entropy and mutual information are introduced and MERMAID (Minimum Renyi’s Mutual Information), an algorithm for ICA based on such these definitions, is here described, implemented and tested over a popular BSS problem in bio-electromagnetism: fetal Electrocardiogram (fECG) extraction. MERMAID was compared to the well known algorithm INFOMAX and it showed to better learn from data and to provide a better source separation. The extracted fECG signals were finally post-processed by wavelet analysis.

The paper deals with an approach that automates the computation of frequential characteristics of passive electronic circuit and their associated sensitivities according to all the components of the circuit. This method enables the designer to focus on the physical behavior of the circuit since the modeling and computing tasks are automatically performed without any computer science skills. It is useful to size circuits with many constraints by using optimization based on gradient algorithm.

This paper proposes an original approach for detection, localization and quantification of faults appearing in electrical machines. The used method in this work deals with the analysis of the leakage magnetic field of a machine. This approach is already known, but until now, classical methods only detect if a fault is present or not. Thus, we propose a new approach based on the theory of inverse problems. It not only enables us to identify a faulty mode, but also to discriminate several types of defects, to localize them and also to quantify their importance.

The resolution of Inverse Problems, especially those resulting in the Medical Diagnostics, is usually difficult because of the inherent noise and inaccuracies present in the data used for the reconstruction. A typical example is given by the Electrical Impedance Imaging, used for the long-term monitoring of “anomalies” present in patients’ bodies. The adoption of soft computing schemes, thanks to their intrinsic capability of dealing with data affected by inaccuracies, reveals effective in this field. As an example, the use of Artificial Neural Networks is proposed here to reconstruct the evolution of a liver tumor treated with thermal ablation.

To catch some data including informal knowledge, the groupware named GUNGEN-SECI has supported an idea generation method with chat conversation data as same as idea data from a brainstorming session. The method with the chat data had some issues; the 70 percent of the chat data were not available and these chat data were not necessary always for the method but has a probability to stimulate human thinking. In this paper, a data flow interface is proposed to support selection and awareness of chat data for the idea generation. The interface makes some flows of the chat data like conveyor belt sushi, which is popular in Japan. The flow limits selection time for handling many chat data and the concurrent usage of the interface with the method is expected to occur an emergence of an idea data.

There are many systems to provide sightseeing information using the Internet, but there are few systems to support the lifecycle of an historic tour. We developed an historical tour support system that provides its users with appropriate information at each phase of the tour. Our system has interesting displays such as 3D graphics and multilingual information for use before sightseeing. This system also provides on the spot information about historic sites using a cellular phone and GPS. After the tour, a user can enjoy a picture album that has been produced in the server during the sightseeing. Twelve tourists tried this system and found it useful.

In recent years, online distribution of academic lectures to distance locations has occurred through the rapid development of Internet technology. However, the main challenge to distance learning is student motivation, since it is more difficult to maintain long-term concentration in an isolated environment. Recently, there are many students who can’t write proper documents. In this circumstance, they have to learn the systematic decision process of specifications and programming using software engineering. We become an author engaged in software development in a corporation, and develop a program. This system can be systematically learned from system designs to programming. The problems will become clear when they make programs with documentation. Firstly, students grasp a design image from general program learning. They learn system design and the documentation necessary to create quality work. Then, this system is used by the Perl language on the Web. The design and program then become compatible on the Web. They study at their own pace by independent learning - provided it is user-friendly. Using this system, the teacher could track and evaluate a student’s degree of understanding, and send individual advice via e-mail. But there is the problem that a frame won’t be designated because it doesn’t decide uniformly. It selects a way of effectively grouping a person and the system.

Although many communication ways exist in a home, it waits these existing communication ways for the network which ties all up. It is necessary to have three important performances in this network. They are a low price, construction needlessness, and a high communication performance. In order to realize this, we are advancing examination of the mutual complement network system which works two communication systems simultaneously. That is, they are wireless communi-cation (IEEE 802.15.4 Zigbee) and wired communications PLC (Power Line Communication). This network has been evaluated at the general home or the comparatively small-scale business building. And since the system which raises a communication performance from this time was obtained, it reports.

In a wireless sensor network (WSN) system, an important problem for the special application, in which the battery of a sensor node cannot be exchanged, is how to extend the life time of the entire WSN system. In the past we proposed an improvement to the LEACH protocol to lengthen the lifetime of a WSN system and showed the superiority of our proposal by simulation. In this paper we set up an experimental WSN system using sensor nodes called “MICAz”. We try to measure the lifetime of the WSN systems based on the protocol of our proposal and to compare with the simulation result. We confirm by experiment that the lifetime of our proposal is up to 1.2 times longer than that of the LEACH protocol at one sensor node. The protocol of our proposal is sufficiently effective though the effect of the simulation cannot be achieved.

In this article, we discuss the evaluation of an impression displayed by presentation contents which are constructed by a dialogue between embodied characters. There are many types of presentation style to provide information to users to get a better understanding. If the way which information is shown through an embodied characters’ dialogue is adopted, the characters have the faces and bodies in order to be ”embodied”, so, the information provider must care about meanings of a state of nonverbal expressions of the embodied characters. People display and exchange nonverbal expressions including eye-gazes, noddings, and facial expressions in daily conversation. Nonverbal expressions convey various kinds of information that is essential to make our face-to-face communication successful. In the previous work on social psychology, it is known that there are interdependences among nonverbal expressions between those from different persons in conversation with each other. We apply this knowledge to a dialogue between embodied characters, which provide information to users, so that we evaluate if the presentation dialogue expresses the corresponding atmosphere to an expected atmosphere.

In many mission-critical systems, databases are updated by transaction processing, and are queried by batch processing to make statistics and so on. It is necessary that both processing can be executed at the same time for the efficient operation. The method, which manages the history of the data and queries a snapshot of the past time to avoid the influence of the present updating, was proposed. However, there is a problem that even if the error data had been corrected just before querying, it is not reflected in the snapshot. In this paper, we propose a method, by which the correction of the data done after the time of the snapshot is reflected in it, and show the integrity of the querying result maintained even while data are being updated. Moreover, we applied this method to a mission-critical system, and confirmed that it was effective.

This paper describes a characteristic of groups in the information systems design class in which a web-based support system was utilized. The groups were analyzed according to the frequency of the discussion within a group on Bulletin Board System (BBS). We observed significant differences among groups in the use of BBS. In an effective BBS collaborative group, members incorporated feed-back from professor and discussed each other’s opinions.

Advanced cooperative work needs user context knowledge in spatial and temporal dimensions. The always-on property of the mobile Internet enables further extension of the cooperative work. It needs to extend the temporal knowledge of the user behaviors for this purpose. This paper explores the temporal dimension: different end-user behavior parameters in different time-scales using the mobile clickstream. The Markov model-based estimation in sub-day scale, day-scale and week-scale transition patterns is analyzed from monthly mobile clickstreams and hierarchical clustering is performed with the three different time-scale behaviors.

We propose a butler-type personal service model for new enterprise value creation. Butler-type personal service is stand on the concept of user-oriented perspective. We apply the butler-type service concept for ubiquitous healthcare service, and model a ubiquitous healthcare service and its provider (ubiquitous healthcare service provider :UHSP).In addition, we concluded that the butler-type service model is effective and suitable for ubiquitous healthcare services.

In order to improve the home network communication performance, research with two different communications, wired and wireless, is catching researchers’ attention. We call this type of network as a mutual complementary home network. In this paper, we discuss some reliable communication methods for the mutual complementary home network and how to implement the methods. We developed a prototype system and evaluated the performance of one method.

Reprogramming is an important service for wireless sensor network to faciliate management and maintenance tasks. It uses the pipelining method, in which code images are divided into several segments, and segments are distributed in parallel. It is an effective way to reduce completion time. As we increase the number of segment divisions, we can increase the speed of code distribution. However, control messages increase as the number of divided segments increase, and that consumes more energy. The relationship between the speed of distribution and the number of control messages is therefore a trade-off. In existing pipelining, the number of segment divisions is determined as an entire network, that but this determination disregards the status of each node. It impairs the nodes which want to reduce the number of messages for various reasons. To solve this problem, we propose the state-based pipelining that can take the state of each node into consideration. In this method, each node can determine the number of segment divisions by itself.

Recommendation increases its importance in research and industries as the e-commerce penetrates every-day life. The authors propose a recommendation method using customers and goods attribute matching. The proposed method is based on Analytic Hierarchy Process (AHP) and Conjoint analysis with purchase records. The authors describe the details of attribute weight adjustment using a mathematical model. We applied it in research experimentally in a real shop and examined practical use.

We have developed an input method that uses sense of touch to input face marks. Using this method, we can input face marks if we grip the mouse. The feature of this system is that we can input two or more kinds of face marks according to how to grip a mouse. We applied this input method to the contents assessment of broadcasting and evaluated the system. We found good evaluation of the face marks that present positive impressions, comparing to the former application.

This paper describes a learning environment that provides program division patterns for programming education. Receiving source codes drawn by learners, this environment analyzes data flow and detects the information on active variables based on lexical, syntax, and semantic analyses. Then, this environment generates and evaluates module division patterns and finally shows visually to a learner module division patterns, active variables and their active scopes.

With the increasing popularity of e-learning, much attention has been focused not only on individualized learning but also on collaborative learning in where several learners participate. In this research, a system designed to actualize collaborative learning is developed. The contents of collaborative learning to run on the system are prepared and subsequently validation experiments have been conducted. This paper describes the agent-oriented collaborative learning system and the contents of the collaborative learning exercise, and reports the results of the validation experiment that is conducted.

Recently the amount of personal contents which a person has to manage is increasing remarkably. We are working on Experience Web, which is a system for managing personal contents based on experiences of a user. Experience Web provides functions to integrate various types of personal contents in a unified manner, and enable a user to search them based on context. Personal contents represent activities of a user and are taken in an activity of a user. Personal contents are used more efficiently if they are shared in the users who attended the same activity. In this paper, we discuss the search that using personal contents of other users.

The authors developed a CSCL environment to utilize “Six Thinking Hats”. This environment is an extension of online discussion room, and extended to assign “hat” to each learner, facilitate to think based on his “hat”,and visualize which hat is on. With use of this environment, an instructor is ableto introduce, facilitate, and train students “Six Thinking Hats” method in onlinelearning environment with less instruction workload compared with face-to-face learning environment. The authors also evaluated the proposingenvironment, and found significant difference of number and quality of ideasfor given problems between “Six Thinking Hats” environment and simple online discussion environment.

The know-how of critical thinking is tacit and latent, and the importance of training critical thinking skills is widely recognized. For those reasons, it is difficult (i) for learners to develop meta-cognitive skills, (ii) for teachers to externalize their tacit know-how of critical thinking, and (iii) for system developers to embed active learning support functions into the system that follow a learner’s thinking processes. To support meta-cognitive learning, CSCL is a promising approach, although we must design functions well to facilitate learner and teacher behaviors (communications). In this paper, we propose a learning-phase model that specifies learning goals and scaffolds for respective learning phases. Furthermore, we illustrate a development system for a nursing domain according to the model. We expect that the model-based system will contribute to encouraging meaningful learning behaviors.

Learners generate their own answers by exchanging their opinions through communication in collaborative learning. Therefore, the conversation record and their private descriptions often include effective information for solving exercises. Such information can be used for assisting other learners who try to solve the same exercise. Our objective is to extract effective knowledge from collaborative learning record and to arrange such knowledge from viewpoint of solving the exercise progressively. Participants’ private descriptions are generated according to the utterance, so that the descriptions can be linked to the effective utterance that is newly uttered. Therefore, in our approach, knowledge is extracted based on participants’ annotations and described time. Then, the discussion knowledge graph is introduced from the conversation record and participants’ descriptions. Learners can find hints easily and solve exercise effectively with discussion knowledge graph.

Physical activities beginners encounter difficulties in being accustomed to their new level of activity for lack of constant motivation in regular physical activities. In order to keep a high level of motivation, the selection of a proper training partner is of significant importance. At present, SNS (Social Networking Service) provides a perfect opportunity to successfully find such an appropriate partner. This paper proposes a methodology for organizing an online community based on the real world log. During training, users will have tiny loggers on, such as GPS (Global Positioning System) and/or HRM (Heart Rate Monitor). After training, users will send the log file from the hardware devices to the system. The system aims at automatically organizes the appropriate community to increase or keep their motivation. This paper reports the design and development of the referred system.

A factor to be considered in browsing of an existing web page to a mobile terminal is the difference in hardware environments between average PCs and terminals. It is necessary to service contents effectively in accordance with diverse environmental information of various terminals caused by the rapid development in communication devices. This various terminal information needs much time to generate the content for a server and much capacity to load in the server. Therefore the method to minimize response time and server capacity is required. This paper proposes a CA(Content Aware Selecting) and HCT(Hyperlink Based Content Transcoding) method to provide a more rapid response time according to requirements of various terminals and illustrates the results of test system.

Since users want to have proper services in their own position and surrounding circumstance in ubiquitous computing, it is critically important to provide appropriate services by adapting changes of the user requirements and their environments. Most approaches lack considerations for position or preference of users. In this paper, we present an adaptation process model to dynamically select services based on user’s preferences according to their contextual changes. We define a conceptual context model and use the adaptation process model to select proper services. The adaptation process consists of the service selection and the service negotiation. Service candidates are selected based on user’s and service provider’s preferences in the service selection process. The service negotiation process determines the best service among service candidates.

Bayesian Networks are proven to be a comprehensive model to describe causal relationships among domain attributes with probabilistic measure of conditional dependency. However, depending on a context, many attributes of the model might not be relevant. If a Bayesian Network has been learned across multiple contexts then all uncovered conditional dependencies are averaged over all contexts and cannot guarantee high predictive accuracy when applied to a concrete case. We are considering a context as a set of contextual attributes, which are not directly effect probability distribution of the target attributes, but they effect on “relevance” of the predictive attributes towards target attributes. In this paper we use the Bayesian Metanetwork vision to model context-sensitive feature relevance. Separating contextual and predictive features is an important task. In this paper we also consider three strategies of extracting context from relevant features, which are based on:

part_of

context, role-based context and interface-based context.

The progress of industrialization and civilization accelerates the complexity of traffic system. To solve the problem of increase of traffic volume and complexity of traffic system, the methods that offer real-time traffic information to drivers like Intelligent Transport System(ITS) are proposed and are researched. Also navigation system that can use in a car is being studied. This paper suggests the selection method of route for the driver’s assistant system that can become individual system to driver by addition of driver’s tendency. Driver’s tendency defines as characteristic of the driver’s driving pattern and the selected driving route. This paper infers driver’s tendency and characteristics of routes by use of fuzzy logic and simulates the proposed algorithm with Personal Computer(PC) and personal Digital Assistant(PDA).

This work presents a clustering method which can be applied to relational knowledge bases. Namely, it can be used to discover interesting groupings of semantically annotated resources in a wide range of concept languages. The method exploits a novel dissimilarity measure that is based on the resource semantics w.r.t. a number of dimensions corresponding to a committee of features, represented by a group of concept descriptions (discriminating features). The algorithm is an adaptation of the classic

Bisecting k-Means

to complex representations typical of the ontology in the Semantic Web. We discuss its complexity and the potential applications to a variety of important tasks.

Ontology mapping is a precondition to achieve interoperability between agents or services using different ontologies. Today, a series of methods for ontology mapping have been reported. However in the existing literature, the focus has so far been on computing similarity and establishing equal mapping relation. In practice, it is possible to have other mapping relations besides equal mapping, such as subclass mapping relation. In this paper a new method for ontology mapping, FCA-Mapping, is presented. FCA-Mapping is based on Formal Concept Analysis (FCA) theory. By computing relation measures between entities of different ontologies, it can automatically establish equal and subclass mapping relations.

This paper presents a distributed cognitive architecture suitable for Ambient Intelligence applications. The key idea is to model an intelligent space as an ecosystem composed by artificial entities which collaborate with each other to perform an intelligent multi-sensor data fusion of both numerical and symbolic information. The semantics associated with the knowledge representation can be used to aid intelligent systems or human supervisors to take decisions according to situations and events occurring within the intelligent space. Experimental results are presented showing how this approach has been successfully applied to smart environments for elderly and disabled.

Due to a limited exposition to the language in its spoken form in their first years of life, deaf children lack the primary means of acquiring literacy skills. In this paper, we present

LODE

a web-based interactive tool for the literacy of Italian deaf children.

LODE

proposes written e-stories and elicits children to globally reason on them through interactive exercises, developed with the support of an automated reasoner and a natural language processor. The

LODE

’s users are also invited to collaborate and exchange their productions in an interactive manner.

Despite the success in the application of ontologies as tools supporting semantic based approaches in various application areas, the lack of simple instruments for their visualization and editing still hinders their diffusion and wide adoption. This paper introduces NavEditOW, a system for the visualization, navigation, updating and maintenance of ontologies through the web. The paper describes the functionalities offered by the system, as well as its internal architecture. A description of its application to the representation and management of archaeological knowledge for the description of publications in an e-library ends the paper.

The present generation of Web browsers provides minimal support to assist users in browsing, making queries and checking results. Although many web sites tried to make their interfaces user friendly and easy to manage, there are no or little ones that ensure the user interface to provide high quality in use for all users who still find difficulties to find what they are looking for on the Web. In this paper, we address this problem by introducing a methodology for designing interfaces of tourism Web portals that adds personalization facilities. Our methodology reflects both the easy use of the Web and the adaptability to the user’s characteristics and preferences. To do this, we investigate the use of the Semantic Web and the Web Usage Mining in order to help a user to explore an information space. To validate the methodology, a semantic search portal tool was designed and is on the way to be implemented as a part of a tourism portal for the Eiffel project.

Semantic-based service discovery is a major open challenge towards a wide acceptance of Semantic Web services. In this paper we present a prototype which implements a fully-automated matchmaking methodology for discovering (compositions of) Web services, described in OWL-S, capable of satisfying a given client request.

This paper presents an approach for the recommendation of items represented by different kinds of features. The motivation behind our research is that often, in online catalogues, items to be recommended are described both by textual features and by non-textual features. For example, books on Amazon.com are described by

title

,

authors

,

abstract

, but also by

price

and

year of publication

. Both types of features are useful to decide whether the item should be recommended to the customer. We propose an approach which integrates non-standard inference services and a Naïve Bayes profiling system able to analyze the textual features of the items by advanced natural language processes and to learn

semantic

user profiles exploited in the recommendation process.

The aim of this work is to introduce a semantic integration between an ontology and a chatbot in an Intelligent Tutoring Systems (ITS) to interact with students using natural language. The interaction process is driven by the use of a purposely defined ontology. In the ontology two types of conceptual relations are defined. Besides the usual relations, which are used to define the domain’s structure, another type of relation is used to define the navigation schema inside the ontology according to the need of managing uncertainty. Uncertainty level is related to student knowledge level about the involved concepts. In this work we propose an ITS for the Java programming language called TutorJ. In our system a chatbot module manages the dialogue in a semantic way. It is capable to deal with the ontology, and also with a LSA-engine. Latent Semantic Analysis (LSA) technique is used to analyze the correctness of the student sentences to establish what concepts she knows. The whole system is explained and the attention is focused on the process for the creation of the correspondence between ontology concepts and student’s answers.

The appearance of powerful tools for lightweight metadata creation, such as collaborative tagging systems, is harnessing the power of online communities, although such metadata are limited to human consumption only. In this paper we first propose an abstract model for representing a generic collaborative tagging system which uses RDF as the underlying technology to store metadata created by different online communities. Then, we present a scenario with the purpose of illustrating how a service able to retrieve tags from different folksonomies can support users in the organization of their personal information spaces within the context of a digital library.

Study on semantic search can be mainly divided into two kinds: one is augmenting traditional keyword-based search engines with semantic techniques like ontology, semantic inference; the other is proposing methods for directly searching on semantic repositories as RDF or OWL files. However, compared with those theoretic-driven approaches, semantic search still lacks applications with real life cases at the moment. By analyzing the scenario of domain resources, this paper proposes a domain-oriented semantic search model, which provides both entity search (

concepts

and

instances

) and relationship retrieval (

concept2concpet

,

concept2instance

and

instance2instance

). The search model can be further refined as a generalized and domain independent architecture with inference supports. The supporting semantic inference is gradually achieved by reasoning-rules, formulas and algorithms. With such semantic search model, more exact and meaningful information hidden in the resource repository can be extracted.

This paper presents the idea to reason over user’s tags to define and enrich the user model. We apply our approach to an adaptive web-based and multi-device social recommender system: iCITY, which exploits a tag-based user model, enriched from the information derived from the tags inserted in the system by users, and filled also with the tags the user has already exploited in other social web sites. Moreover, we propose an architecture to enable the iCITY tag-based user model be exported and shared with other social applications in a semantic enhanced way. Finally, we propose the sharing of the user profile, together with the list of tags, in a shared syntax (such as RDF(S), OWL, RSS).

This paper proposes the SoundSpot system, a location/user-dependent audio guide system to reduce many restrictions of existing audio-based annotations. It can track the positions of visitors and then only provide sound information to the limited spots around them. We experimented it in an elementary school, where students listened to annotations played by it and their understanding was evaluated. The result of the questionnaire based survey and their answers to the quiz regarding the contents of the sounds proved that the system had the possibility to enable users to enough listen to and understand the contents.

This paper describes a discussion support function on the intelligent bulletin board system (iBBS), which aim to enhance educative effectiveness of e-Learning programs. Bi-directional communication plays an important role in asynchronous e-Learning systems. The authors have proposed the support functions for activization of communication. The paper proposes implementation and improvement for the iBBS, and demonstrates its performance through an experimental evaluation.

In Web-based collaborative learning, it is hard for learners to exchange opinions because communication means are too limited. Therefore, learners cannot be immersed in the learning environment. In order for learners to collaborate with others effectively, they need be able to acquire information on the situation, the behavior of others, etc. and feel the atmosphere of the learning environment by grasping the intentions of others. In this paper, we propose an immersive round-table interface in which learners’ intentions for focusing on others are reflected by automatically changing camera images of other learners based on the focus degree. Based on the experimental results, the concept of the round-table interface clearly helps learners concentrate on the learning environment and communicate with others naturally.

We have developed

AirTransNote

, a student notes sharing system to facilitate collaborative and interactive learning in a regular lecture at conventional classrooms. Our former student system employed ultrasonic digital pens to capture student notes on a usual paper sheet. Although the paper-based approach is intuitive for learners, the ultrasonic pen with a PDA created some difficulties when used by senior high school students. In order to eliminate those difficulties, we have introduced anoto-based pens and a data gathering system called “digital pen gateway.” Owing to the improved student system, students could easily submit their answers with handling of multiple paper sheets. The teacher could refer the answers to choose students who need assistance during the lecture. Questionnaire results showed that the simpler student interface is quite acceptable, and the system can realize our augmented classroom concept in a practical way.

Meta-cognition plays an important role in acquiring expertise and transferring it. We recognize the necessity of building a learning scheme for developing meta-cognitive skills but much knowledge for it has not been acquired. The reason is that it is difficult for learners and instructors to discuss meta-cognitive skills as learning topics since the knowledge for performing meta-cognitive skills is quite tacit, latent and context dependent. Our goal is to build a new learning scheme to support meta-skill learning. In this paper, we firstly describe our basic idea on how to reify meta-cognitive skills as learning topics. We secondly clarify that the presentation task is useful for dealing with meta-cognitive skills as learning topics. Then we finally design a CSCL environment for building meta-cognitive skills.

For the equations of population dynamics, this note presents three symmetries: a coordinate symmetry, an additive symmetry and an exchange symmetry. Among them, additive symmetry is a new one that should be held in equations of population dynamics particularly those for quasispecies. The distinguishability between species is also stressed to obtain the stability condition of 2-dim Lotka-Volterra model that satisfies the additive symmetry.

This work presents an analysis of the static Aging operator for different evolutionary algorithms: two immunological algorithms (OptIA and Clonalg), a standard genetic algorithm SGA, and Differential Evolution (DE) algorithm. The algorithms were tested against standard benchmarks in both unconstrained and dynamic optimisation problems. This work analyses whether the aging operator improves the results when applied to evolutionary algorithms. With the exception of DE, the results report that every algorithm shows an improvement in performance when used in combination with Aging.

We contrast properties and parameters of a Pareto distribution law with the behavior of memory endowed processes underlying the intercontact times of opportunistic networks. Within a general model where mobile agents meet together as a consequence of a common goal they are carrying out, the memory of the process identifies with the agent intention versus a goal, where intention consists in turn in the introduction of asymmetries into a random walk. With these elementary hypotheses we come to a very elementary agents mobility model as a semantic counterpart of the Pareto law. In particular this model gives a suitable meaning to law parameters and a rationale to its fitting of a benchmark of real intercontact times.

This paper focuses on anomaly detection in user behavior. We present a review of our immunity-based anomaly detection system, and propose a framework of the immunity-based anomaly detection system with a new mechanism of diversity generation. In the framework, each computer on a LAN generates diverse agents, and the agents generated on each computer are shared with all other computers on the LAN. The sharing of agents contributes to their diversity. In addition, we propose an evaluation framework of immunity-based anomaly detection, which is capable of evaluating the differences in detection accuracy between internal and external malicious users.

Genomic DNA copy number aberrations are frequent in solid tumours although their underlying causes of chromosomal instability in tumours remain obscure. In this paper we show how Artificial Immune System (AIS) paradigm can be successfully employed in the elucidation of biological dynamics of cancerous processes using a novel fuzzy rule induction system for data mining (IFRAIS) [1] of aCGH data. Competitive results have been obtained using IFRAIS. A biological interpretation of the results carried out using Gene Ontology is currently under investigation.

As a theory on a deciding factor leading to AIDS, ”antigenic diversity threshold theory” has been offered by Nowak et al. This theory mentions AIDS develops when a number of mutant strains from a single HIV swells through mutations over a fixed critical number. Mutation is mainly due to transcription errors by a reverse transcriptase. Existing HIV models assume the transcription error probability (i.e. mutation probability of HIV) is constant. However it being considered the reverse transcriptase builds its gene into HIV genes as its operand, it would be natural to guess the transcription error probability varies at each transcription. Hence, this study proposes a HIV dynamical model factoring into the mechanism of HIV mutating its mutation probability. Along with earlier studies this study discusses the subject of the antigenic diversity threshold and demonstrates the proposed model has a threshold with the minimal mutation probability influencing development of AIDS.

The complex network theory has lately drawn considerable attention, because it can successfully describe the properties of many networks found in nature. In this paper, we perform the immunity-based diagnostic model on complex networks, namely, small-world networks and scale-free networks. For the distributed diagnosis, the number of nodes on large-scale networks has little effect on the diagnosis capability. In addition, some results show the performance of the diagnosis model depends on the number of links and the average path length. The results can give us valuable knowledge to incorporate the immunity-based diagnosis into real complex networks.

Antigenic diversity threshold theory by Nowak and May proposes there will be a threshold of diversity for HIV strains in a population dynamics with asymmetric interaction between specialist (T-cells) and generalists (HIVs). This paper revisits the threshold condition from a symmetric point of view. We consider how stability condition for HIVs will be altered when the asymmetry of specialists/generalists is relaxed (made symmetrical). Diversity measure is also considered from a symmetric point of view involving a distance between parameters characterizing strains.

In the paper a new Multi-Layers approach (called Multi-Layers Model

MLM

) for the analysis of stochastic signals and its application to the analysis of gene expression data is presented. It consists in the generation of sub-samples from the input signal by applying a threshold technique based on cut-set optimal conditions. The

MLM

has been applied on synthetic and real microarray data for the identification of particular regions across DNA called

nucleosomes

and

linkers

. Nucleosomes are the fundamental repeating subunits of all eukaryotic chromatin, and their positioning provides useful information regarding the regulation of gene expression in eukaryotic cells. Results have shown a good recognition rate on synthetic data, moreover, the

MLM

shows a good agreement with a recently published method based on Hidden Markov Model when tested on the

Saccharomyces cerevisiae

chromosomes microarray data.

In this work we present an integrated set of tools allowing a multi-step process that, starting from raw datasets, brings them through dimensionality reduction, preclustering analysis and clustering assessment, to a visual and interactive environment for data exploration. At the core of the process lies the idea of subdividing the process of data clusterization into different steps: a preliminary analysis in which algorithmic parameters are estimated, a clustering step based on the previous analysis and, finally, a clusterization assessment step including interactive clustering. This last step allows users to participate in the process of clustering and helps them figuring out the data underlying structures. The models are actually implemented in a group of integrated, user-friendly tools under the MATLAB environment, featuring a number of classical and novel data processing, visualization, assessment and interaction methods.

Assessing the quality of a clustering’s outcome is a challenging task that can be cast in a number of different frameworks, depending on the specific subtask, like estimating the right clusters’ number or quantifying how much the data support the partition given by the algorithm. In this paper we propose a computational intensive procedure to evaluate: (i) the consistence of a clustering solution, (ii) the informativeness of each feature and (iii) the most suitable value for a parameter. The proposed approach does not depend on the specific clustering algorithm chosen, it is based on random permutations and produces an ensemble of empirical probability distributions of an index of quality. Looking to this ensemble it is possible to extract hints on how single features affect the clustering outcome, how consistent is the clustering result and what’s the most suitable value for a parameter (e.g. the correct number of clusters). Results on simulated and real data highlight a surprisingly effective discriminative power.

Searching for structures in complex bio-molecular data is a central issue in several branches of bioinformatics. In particular, the reliability of clusters discovered by a given clustering algorithm have been recently assessed through methods based on the concept of stability with respect to random perturbations of the data. In this context, a major problem is to assess the confidence of the measures of reliability. We discuss a partially ”distribution independent” method based on the classical Bernstein inequality to assess the statistical significance of the discovered clusterings. Experimental results with gene expression data show the effectiveness of the proposed approach.

The application of various clustering techniques for large-scale gene-expression measurement experiments is an established method in bioinformatics. Clustering is also usually accompanied by functional characterization of gene sets by assessing statistical enrichments of structured vocabularies, such as the Gene Ontology (GO) [1]. If different cluster sets are generated for correlated experiments, a machine learning step termed

cluster meta-analysis

may be performed, in order to discover relations among the components of such sets. Several approaches have been proposed for this step: in particular,

kernel methods

may be used to exploit the graphical structure of typical ontologies such as GO. Following up the formulation of such approach [2], in this paper we present and discuss further results about its applicability and its performance, always in the context of the well known Spellman’s Yeast Cell Cycle dataset [3].

The data representation strategy termed “Membership Embedding” is a type of similarity-based representation that uses a set of data items in an input space as reference points (probes), and represents all data in terms of their membership to the fuzzy concepts represented by the probes. The technique has been proposed as a concise representation for improving the data clustering task. In this contribution, it is shown that this representation strategy yields a spectral clustering formulation, and this may account for the improvement in clustering performance previously reported. Then the problem of selecting an appropriate set of probes is discussed in view of this result.

Agent-Oriented Software Development is one of the recent contributions to the field of Software Engineering. To date numerous methodologies for agent-oriented software development have been proposed in the literature. However, their application to real-world problems is still limited due to their lack of maturity. Evaluating their strengths and weaknesses is an important step towards developing better methodologies in the future. This paper presents research results obtained by applying three agent-oriented methodologies, namely ROADMAP, MaSE and Prometheus in the context of an E-commerce system. The results are presented and future work is discussed.

RFID technology can be found embedded in almost everything from razor blade packages, clothing and books to prescription medicines to parts of an aircraft. Despite the pervasive nature of RFID, surveys have consistently shown a lack of RFID awareness, an overall lack of understanding and general confusion about what it actually is, its capabilities and limitations. This is due in part to a lack of a comprehensive, principles-based and systematic RFID classification scheme. This paper proposes a multi-agent architecture for RFID taxonomy of currently available RFID technology and systems. The taxonomy is based on a sound service-oriented and multi-agent RFID architecture framework. Multi-agent plays a role in the formulation of the taxonomy. It can be used by both novices and RFID practitioners to gain an understanding of the “next big thing” technology, the architectural considerations for designing and implementing a successful RFID system.

Radio Frequency Identification (RFID) has recently received much attention from various parties involved in supply chain management. In the RFID system, a reader recognizes tags through communications over a shared wireless channel. When multiple tags transmit their IDs at the same time, the tag to reader signals lead to collision. Various tag identification algorithms have proposed for RFID systems. In this paper a modified Adaptive Binary Splitting (MABS) protocol that is to improve the binary tree protocol and reduce collisions efficiently via multi-agent systems.

Agent coordination and cooperation co-exist in a multiagent system, and are cognitively linked. This link is emphasized further by incorporating their atomic composition with their ability to perceive and gather information from the environment around them. As a result, a new generation of coordination/cooperation architectures is starting to emerge. From each of their definitions and current implementations, we show how the relationship between the Belief-Desire-Intention (BDI) architecture and the Observe-Orient-Decide-Act (OODA) loop can enable the use of Coordinative Cooperation within the Agent Coordination and Cooperation Cognitive Model. More importantly, we show the relationship between coordination, cooperation, BDI and OODA. This paper also discusses the current developments of the model and how the BDI and OODA architectures can affect coordination and cooperation within a Multi-Agent System (MAS). We recommend how these concepts can be designed and implemented in a MAS.

Agent technology has been claimed to be a promising tool for creating e-commerce applications. To develop a multi-agent e-commerce system effectively, the developers have to deal with several issues such as agent characteristics, agent functionalities, protocols, communication, cooperation and coordination. JADE (Java Agent Development Environment), an open source software framework, is currently the state-of-the-art tool used for developing multi-agent systems. In this paper, our aim is to present our experience in designing and developing an e-commerce multi-agent system with JADE. Our objective is realised through the presentation of the analysis, design and implementation phases of an agent software system we currently develop in the context of travel industry. We will also discuss some lessons we have learned in using the JADE platform.

As technology grows, so do the demands on information systems processing. Modern information systems need to handle diverse tasks that tend to be increasingly complex. Automation of processing by employing artificial intelligence techniques is an essential environment for ensuring the smooth running of information systems. In this paper, we investigate the application of multi-agent information systems for online medical surveys. In particular, our contribution is to extend our previous research results to a multi-agent technology for data perturbation as a means of camouflaging database query results. The decision on when to camouflage data is made by a system of collaborating intelligent agents. When a user query request is entered into the system, processing rules are applied to determine whether the result of the query needs to be camouflaged.

UMTS provides a high mobility two-way multimedia service with a medium bit rate. DVB provides high bit rate mobile reception, but restricted to a unidirectional one-to-one. The convergence of telecommunication and broadcasting networks have problem about how to support mutual authentication between user and broadcasting network. In this paper, we propose a new authentication scheme that authenticates each other. We involve UMTS AKA to share session key between user and converged networks. The AKA is based on symmetric keys, and runs typically in a UMTS subscriber identity mobile, a smart card like device. AKA mechanism is based on a challenge and response authentication protocol conceived in such a way as to achieve maximum compatibility with UMTS subscriber authentication and key establishment protocol.The VLR/SGSN checks user identity such as IMSI with PSI/SI which is transmitted through interface

I

i

. If the UMTS authentication user corrects PSI/SI table, DVB-H network transmits data with unnecessary authentication. The session key is used in every session to encapsulate the data sent to the user. Our proposed authentication scheme provide a fast and secure authentication for the converged networks.

Several content distribution services via the Internet have been developed, and a number of bidirectional broadcasting services will be provided in the near future. Since such bidirectional broadcasting treats personal information of the users, provider authentication is necessary. Taking the currently existing broadcasting system using CAS cards into account, Ohtake

et al.

recently proposed the provider authentication system which utilizes key-insulated signature (KIS) schemes. However, the authors did not refer to details of what kind of KIS should be used. In this paper we supplement their works in terms of KIS specification. We carefully identify what kind of KIS should be used and propose concrete KIS schemes which realize both the reliability and the robustness required for the bidirectional broadcasting service.

In bidirectional broadcasting services via networks, all user operations and actions are controlled with the user ID and are recorded by the service providers. This means the user’s personal information such as his/her payment method and preferences is disclosed to the providers possibly against the user’s will. Hence, to preserve privacy, an anonymous authentication scheme for subscription services would seem to be in order. In this paper, we introduce a subscription service model and its system requirements for preserving privacy. Moreover, we propose an efficient group signature scheme having a property of linkability, which is useful for anonymous authentication and a preference subscription service. We show that our authentication scheme is efficient in terms of signature size and that it is suited to subscription services having many members.

When subscribers watch pay-TV programs, they get keys for decrypting the transmitted content and they are charged a fee, so these schemes enable content providers to get information about the contents that the subscribers watched, and the anonymity disappears. Recently, people have been taking the preservation of privacy seriously, and they do not want their private information to be disclosed even to the content provider of pay-TV. Moreover, content providers require a secure and correct revenue sharing scheme. In this paper, we propose an

anonymous pay-TV system with secure revenue sharing

to meet such requirements.

Several new video and image watermarking proposals are based on

Informed Coding

and

Informed Embedding

. However, these systems can be not easily used in fingerprinting schemes because they do not satisfy the marking assumption defined in [1]. In this paper we discuss some guidelines to adapt a watermarking system based on informed coding and informed embedding to a generic fingerprinting code, while keeping up with the marking assumption, that is to say, when as a result of one collusion attack of two users, that have different marks that represent the value 0 in the

n

th position, we have a pirate mark wich represents the 0 value in this same

n

th position. This can be achieved modifying the work of Miller, Doër and Cox in [2].

Personalization of user interfaces (UI) for browsing content is a key concept to ensure content accessibility. In this direction, we introduce concepts for representing end-user requirements that result in the generation of personalized user interfaces for browsing XML content. Representation of these requirements and the process for generating UI are described and illustrated using a case study. With the emergence of the semantic Web and connected XML applications, such personalized user interfaces can be useful for many kinds of users.

Two fuzzy models for real estate appraisal, i.e. Mamdani-type and Takagi-Sugeno-Kang-type have been built with the aid of experts. Both models comprised 7 input variables referring to main attributes of a property being appraised. In order to determine the rule bases for both models an evolutionary algorithm has been applied. The experiments revealed that models assured acceptable estimations of property values.

The typical content-based recommendation systems make use of textual similarity between items. Based on the knowledge about historical user behaviours extracted from the web logs, the content recommendation lists can be verified and filtered: some items are reinforced whereas some other are weakened. Four different usage patterns are used in the filtering process: positive and negative association rules, positive sequential patterns and negative sequential patterns. The last ones are the new pattern concept introduced in the paper.

The current Information Retrieval Systems return hundreds or thousands of documents in response to a query. Users consider only the first 20 or 30, but documents are often sorted according to the query and the results are perhaps not relevant to the users’ needs. This situation is even more problematic when non expert users have difficulties in expressing their information requirements. One solution to this problem could be the use of a user model which would complement the query in order to find the best solutions for the user. Most personalized retrieval systems have a single user model for each user which works on the basis that this user will have always the same information needs. Our proposal however is a personalization method based on the current objective of the user. To this end, we have developed two probabilistic algorithms NBM and WNBM which support different kinds of multi parameter databases and in this paper we present new experiments with these algorithms in METIORE to validate our proposal.

In the paper web-based system user interface hybrid recommendation method based on the ant colony metaphor is presented. In this paper we apply the ontology-based user and user interface modeling. The role of ontologies in the user and user interface modeling is determining objects, concepts and relations between them. The user model is represented as a tuple and user interface model is represented by a set of connected nodes. The recommendation is performed using ant colony metaphor for selection the most optimal path in the user interface graph that specifies the user interface parameters for the specified user. The applied hybrid recommendation method is based on: demographic, content-based and collaborative filtering.

PC users can retrieve lots of common information by Internet search engines. A text-to-speech (TTS) system allows citizens to easily access the public report from the city etc. However it takes a long time for them to listen to the voice information. We propose a fast reading spatial knowledge system in order to minimize the listening time and to supplies directional spatial knowledge producing a synergy between a packaging method and ultrasonic beams. We clarify the “magical number” of human memory during high speed listening and propose a packaging method for the minimization. As a result listeners could listen to an explanation 4.8 times as fast as the regular speed of TTS system. We also discuss features and applications of the system.

Recently cars are been equipped with intelligent systems aiding safe driving. However a large percentage of traffic accidents between human and automobile happen on crosswalk. Intelligent functions from road side are also indispensable in order to make crosswalk safer. We propose a quiet navigation system for safe crosswalks by parametric speakers in order to allow pedestrians to safely cross streets for 24 hours and to find an easy way to their destination. In this paper we examine different types of new parametric speakers comparing them with regular traffic signal loudspeakers and propose a useful type of the speakers.

In this study, the effectiveness of pictogram meanings for lathe procedures and cautionary instructions was investigated. Pictograms were created for instruction in lathe work and the comprehensible levels of the pictograms were analyzed through a survey with 108 subjects. The subjects responded to questionnaires with a five-point scale to determine the clarity of the pictograms and the well-recognized emergency exit and toilet signs. T-tests between some of the pictograms and the signs showed no significant differences. In addition, the average scores given to some of the procedural pictograms were higher than those to the escape sign. These results indicate that the pictograms can be used effectively to give instructions to lathe workers in the manufacturing factories where verbal communication is not easy between supervisor and workers due to language barriers.

Context-aware computing that recognizes the context in which a user performs a task is one of the most important techniques for supporting user activity in ubiquitous computing. To realize context-aware computing, a computer needs to recognize the user’s location. This paper describes a technique for location detection inside a room using radio waves from a user’s computer. The proposed technique has to be sufficiently robust to cater for dynamic environments and should require only ordinary network devices, such as radio signal emitters, without the need for special equipment. We propose performing localization by relative values of RSSI (Received Signal Strength Indicator) among wireless nodes, and also our system support the node mobility. We evaluate the performance of our system in the environment where the node is movable.

Skill knowledge is important human knowledge as well as symbolic knowledge. Definition of skill knowledge is knowledge on high level technique that human acquire through repeated training. Skill knowledge is needed for arts, sports and crafts. It concerns human body structure, muscles, and joints. Also, skill knowledge concerns process of human interaction with objects, namely, recognition and action. There are many researches on learning support system for describable knowledge, such as mathematics, physics, grammar, etc. However, there is few pieces of research on learning support system for skill such as arts and sports. This paper analyses skills and generalize skill features. Then we describe problems to diagnose skill errors. After that, we introduce a learning support system for sketch as an example of a skill learning support environment.

Access control for semi-structured data is nontrivial, as witnessed by the number of access control approaches in literature. Recently, a case has been made for expressing access constraints at finer levels of granularity on data nodes and extending constraints to structural relationships. In this paper, we introduce a rewrite-based approach for access constraint enforcement, based on the ACXESS framework we developed at Indiana University. The ACXESS framework utilizes virtual security views and introduces a set of rewrite rules that takes advantage of the Security Annotated Schema (SAS) - an internal representation for virtual views. It is capable of rewriting user queries against security views into queries against the source data, while honoring the access constraints.

Many scenarios require models for access control which consider former accesses to permit or deny access. One such model is the Chinese Wall model (CWM) which was designed to prevent the misuse of insider knowledge in the consulting business. In this paper, we show that our model for history-based access control provides a flexible and expressive method to define access depending on histories, which contain information about former accesses. We demonstrate this by modeling the policies of the CWM in a way that avoids the unnecessary restrictions of the original CWM.

We study the problem of updating XML repository through security views. Users are provided with the view of the repository schema they are entitled to see. They write update requests over their view using the XUpdate language. Each request is processed in two rewriting steps. First, the XPath expression selecting the nodes to update from the view is rewritten to another expression that only selects nodes the user is permitted to see. Second the XUpdate query is refined according to the write privileges held by the user.

Since XML became the core meta language for many data formats, we need a fine-grained access control model for XML to protect sensitive information carried by XML elements or by relationships between these elements. Several models have already been suggested, but we claim that none of them is sufficiently expressive to properly express some basic security requirements, especially those related to entity relationships protection. To cope with these limitations, we suggest to structure the access control policy using the new concept of

block

. This is used to hide relationships between nodes selected in different blocks. It provides means to specify confidentiality restriction associated with some relationships. An access control model, called XML-BB (XML Block Based Access Control), that includes this concept of block is presented and the implementation of this model is described.