Intelligent Information and Database Systems

The paper deals with an approach to image segmentation using interval type-2 fuzzy subtractive clustering (IT2-SC). The IT2-SC algorithm is proposed based on extension of subtractive clustering algorithm (SC) with fuzziness parameter

m

. And to manage uncertainty of the parameter

m

, we have expanded the SC algorithm to interval type-2 fuzzy subtractive clustering (IT2-SC) using two fuzziness parameters

m

1

and

m

2

which creates a footprint of uncertainty (FOU) for the fuzzifier. The input image is extracted RGB values as input space of IT2-SC; number of clusters is automatically identified based on parameters of the algorithm and image properties. The experiments of image segmentation are implemented in variety of images with statistics.

This paper introduces the use of GA with a novel fitness function to eliminate noisy and irrelevant features. Fitness function of GA is based on the Area Under the receiver operating characteristics Curve (AUC). The aim of this feature selection is to improve the performance of

k

-NN algorithm. Experimental results show that the proposed method can substantially improve the classification performance of

k

-NN algorithm in comparison with the other classifiers (in the realm of feature selection) such as C4.5, SVM, and Relief. Furthermore,this method is able to eliminate the noisy irrelevant features from the synthetic data sets.

Associative classification aims to discover a set of constrained association rules, called Class Association Rules (CARs). The consequent of a CAR is a singleton and is restricted to be a class label. Traditionally, the classifier is built by selecting a subset of CARs based on some interestingness measure.

The proposed approach for associative classification, called Associative Classifier based on Closed Itemsets (ACCI), scans the dataset only once and generates a set of CARs based on closed itemsets (ClosedCARs) using a lattice based data structure. Subsequently, rule conflicts are removed and a subset of non-conflicting ClosedCARs which covers the entire training set is chosen as a classifier. The entire process is independent of the interestingness measure. Experimental results on benchmark datasets from UCI machine repository reveal that the achieved classifiers are more accurate than those built using existing approaches.

Clustering is the most important task in unsupervised learning and clustering validity is a major issue in cluster analysis. In this paper, a new strategy called Clustering Algorithm Based on Histogram Threshold (HTCA) is proposed to improve the execution time. The HTCA method combines a hierarchical clustering method and Otsu’s method. Compared with traditional clustering algorithm, our proposed method would save at leastten several times of execution time without losing the accuracy. From the experiments, we find that the performance with regard to speed up the execution time of the HTCA is much better than traditional methods.

Sensor nodes having the limited resource, energy efficiency is an important issue. Clustering on the sensor networks reduces the volume of inter-node communications and raises energy efficiency by transmitting the data collected from members by a cluster head via a sink node. But, due to radio frequency characteristics, interference and collision can occur between neighbor clusters, the resulted re-transmission is more energy consuming. The previous problems occurred between neighbor clusters can be resolved by assigning channels which do not overlap between neighbor clusters. In this paper, we propose a method which assigns and reuses channels which do not overlap between neighbor clusters in dynamic cluster sensor networks. The resource allocation model of the proposed method is analyzed by correctness and simplicity.

Collaborative tagging has become increasingly popular as a powerful tool for a user to present his opinion about web resources. In this paper, we propose a method to generate tag-based profiles for clustering users in a social music site. To evaluate our approach, a data set of 1000 users was collected from last.fm, and our approach was compared with conventional track-based profiles. The K-Means clustering algorithm is executed on both user profiles for clustering users with similar musical taste. The test of statistical hypotheses of inter-cluster distances is used to check clustering validity. Our experiment clearly shows that tag-based profiles are more efficient than track-based profiles in clustering users with similar musical tastes.

In order to aggregate the professional knowledge of examiners (of patent office) in the IPC code assignment and the innovative information within the patent documents, an IPC-based clustering is proposed for formulating the technology strategy. Technology strategy represents managers’ efforts to think systematically about the role of technology in decisions affecting the long-term success of the organization. IPC-based clustering is utilized to generate the technical categories via the IPC and Abstract fields, while link analysis is adopted to generate the relation types for the whole dataset via the Abstract, Issue Date, and Assignee Company fields. During experiment, the technical categories have been identified using IPC-based clustering, and the technology strategies for significant companies have been formulated through link analysis. Finally, the technical categories and technology strategies will be provided to the managers and stakeholders for assisting their decision making.

Cluster Network technologies have been evolving for the past decades and still gaining a lot of momentum for several reasons. These reasons include the benefits of deploying commodity, off-the-shelf hardware (high-power PCs at low prices), using inexpensive high-speed networking such as fast Ethernet, as well as the resulting benefits of using Linux. One of the major difficulty encounters by cluster administrator in the exploitation of Cluster Networks is handling common tasks, such as setting up consistent software installation or removing on all the nodes or particular node and listing of files or processes will require a lot of time and effort and may affect productivity of the cluster. Even though there are numerous systems available which is known as Cluster Middleware (CM), most of it is developed specifically for in-house use such as scientific research or commercial purpose. Some of them mainly design for job execution rather than node management. To mitigate this problem in the cluster network environment, Cluster Control Management (CCM) system has been developed as a solution for this problem. CCM presently contains six tools, and is extensible to allow more tools to be added. All this six tools are designed for remote cluster administration which includes Switch Manager, Program Manager, Report Manager, User Manager and Administrator Manager. In this paper, we discuss the architecture and the design for CCM using a prototype called UKM C

2

M deploys using open-source software (OSS).

This paper presents a novel saliency detection approach using multiple features. There are three types of features to be extracted from a local region around each pixel, including intensity, color and orientation. Principal Component Analysis(PCA) is employed to reduce the dimension of the generated feature vector and kernel density estimation is used to measure saliency. We compare our method with five classical methods on a publicly available data set. Experiments on human eye fixation data demonstrate that our method performs better than other methods.

In this paper, we present a novel motion vector based information hiding algorithm for H.264/AVC against motion vector steganalysis. For resisting motion vector steganalysis, the underlying statistics of the motion vectors used by motion vector steganalysis are remained during the secret information embedding process. We choose one part of motion vectors to be modified for restoring the statistics used by motion vector steganalysis. In order to guarantee imperceptibility, secret information is embedded by modulating the best search points of other part of motion vectors during the quarter-pixel motion estimation process. Experimental results show that the proposed algorithm effectively resist motion vector steganalysis, while having good video quality.

An encoding and decoding method for Multiple System Barcode (MSB) based on QR code is designed in the paper through fusing several Black and White Barcodes (BWB) representing different color planes. The QR code standard is compatible and could be extended to MSB. MSB’s capacity is increased in the method. The method can be applied to other standard barcode and large-capacity data storage and transmission. The experimental results are encouraging and demonstrate that our method is effective which is easy to understand, implement and extend.

A new research method for the sleepy lizards based on the KNN algorithm and the traditional social network algorithms is proposed in this paper. The famous paired habit of sleepy lizards is verified here based on our proposed algorithm. In addition, some common population characteristics of the lizards are also introduced by using the traditional social net work algorithms. Good performance of the experimental results shows efficiency of the new research method.

In this paper, two novel image feature extraction algorithms based on directional filter banks and nearest feature line are proposed, which are named Single Directional Feature Line Discriminant Analysis (SD-NFDA) and Multiple Directional Feature Discriminant Line Analysis (MD-NFDA). SD-NFDA and MD-NFDA extract not only the statistic feature of samples, but also the directionality feature. SD-NFDA and MD-NFDA can get higher average recognition rate with less running time than other nearest feature line based feature extraction algorithms. Experimental results confirm the advantages of SD-NFDA and MD-NFDA.

We present a secret sharing scheme based on absolute moment block truncation coding (AMBTC) and Hamming code theory. AMBTC is composed of bit planes and two representative gray level pixel values, which are the high and low range means. In this secret sharing scheme, a dealer distributes a secret image to the participants as shadow images. It is important in a secret sharing scheme to keep the shadow images confidential. We therefore propose verifying the authenticity of the reconstructed secret image during the revealing and verifying phase. In order to improve confidentiality, our proposed scheme uses meaningful AMBTC compressed images. Experimental results show that the reconstructed secret sharing images are the same as the original secret sharing images and that the proposed scheme exhibits good performance compared to that of previous schemes.

Image inpainting is to remove unnecessary objects or reconstruct damaged parts of an image automatically. In order to reduce the capacity of the file, the video film is usually stored after the quad or the octal processing. It is this processing that causes an image of low quality and makes the moving object become indistinct. In this paper, we proposed a new image inpainting method for rigid moving object in the temporal domain, in which the pixels are patched by the neighboring frames. The more neighboring frames are patched, the better a PSNR of the moving object image is obtained. The experiment results have shown that our method has good performance and obtained a better quality of the rigid moving object.

This paper presents automatic image matting using component-hue-difference-based spectral matting to obtain accurate alpha mattes. Spectral matting is the state-of-the-art image matting and it is also a milestone in theoretic matting research. However, the accuracy of alpha matte using spectral matting is usually low without user intervention. In the proposed method, k-means algorithm is used to generate components of a given image. Next, component classification is used based on the hue difference of components to obtain the foreground, background, and unknown components. The corresponding matting components of the foreground, background, and unknown components are obtained via a linear transformation of the smallest eigenvectors of the matting Laplacian matrix. Finally, only matting components of the foreground and unknown components are combined to form the complete alpha matte based on minimizing the matte cost. Experimental results show that the proposed method outperforms the state-of-the-art methods based on spectral matting.

This paper aims to present the idea of robotics leadership. By investigating leadership definitions and identifying domains where humans have failed to lead, this paper proposes how robots can step in to fill various leadership positions. This is exemplified by referring to two examples, stock brokering and transportation, and explains how robots could be used instead. Furthermore, this paper aims to provoke discussion by identifying firstly some potential limitations of robots in leadership positions and secondly by proposing that our current technological ecosystem not only is suited for machines to assume leadership positions but rather is inherently headed towards it.

Social media have been one of the most popular online communication channels to share information among users. It means the users can give (and take) cognitive influence to (and from) the others. Thus, it is important for many applications to understand how the information can be propagated. In this paper, we focus on social tagging systems where users can easily exchange tags with each other. To conduct experimentation, a tag search system has been implemented to collect a dataset from Flickr.

Social network sites (SNSs) have been drastically increasing in use in recent years and can be cited as a communication tool with diversified forms in comparison with existing media. People are conducting various activities including relaxing entertainments, information sharing, escapism, social interaction, habitual pass time and others through the use of the SNSs. However, people have various motivation when using SNSs, with information sharing drawning a lot of organizational attention. Therefore, this study aims to ascertain motivational factors of SNSs that influence information sharing and conduct empirical analysis based on use and gratification theories and social identity theory. Factors influencing motivation of use of SNSs were divided into self-expression, involvement, interaction and media structure, and analysis was conducted to determine effects on continuance SNSs motivation. Our analysis results show that involvement had the greatest effect on continuance motivation and that remaining factors were also significant. In addition, continuance motivation turned out to have a significant effect on information sharing.

The deployment of business intelligence (BI) involves complex processes of data reconfiguration and resource alignment. This study investigated whether the issues of data environment and profitability affect BI implementation for the manufacturers that have already adopted enterprise resource planning systems. We individually considered the factors of data warehousing, online analytical processing (OLAP), and data mining for the data environment, while return on assets, return on sales, and return on investment were transformed into a single component of profitability using principal component analysis. Through logistic regression, we determined that OLAP and data warehousing play important roles in the adoption of BI; however, data mining and profitability indicated no such influence.

This research investigates the impact of IT on the cost efficiencies and cost frontiers of the insurance industry. In addition, we compare the differences of the effect of IT on the insurance industry across countries. Moreover, we test the international market’s integration hypothesis and explore the influences of other factors (e.g., scope economy, macroeconomic variables, etc.) on the cost efficiencies of the insurance industry. The evidence from both the two-equation and Tobit models indicates that IT improves cost efficiencies for the developed countries but reduces those for the newly developed economies under study. That is, the international productivity paradox hypothesis is rejected for the insurance firms in the developed countries considered. In addition, results of the cost frontiers for the insurance firms suggest that, IT improves the cost efficiency for the developed countries but not for the emerging economies.

The applying of data mining techniques in banking is growing significantly. The volume of transaction data in banking is huge and contains a lot of useful information. Detecting money laundering is one of the most valuable information which we can discover from transaction data. This paper will propose the approaches on money laundering detection techniques by using clustering techniques (a technique of data mining) on money transferring data of banking system. Besides, we present an implemented system for detecting money laundering in Viet Nam’s banking industry by using CLOPE algorithm.

Small microfinance institutions (in underdeveloped countries), despite their determination to expand into rural areas, are limited by geographic isolation and high transaction costs. For this, a solution has been proposed to access some features of the enterprise application via Smartphone. This solution is based on a service-oriented architecture in which the main features are developed as web services. The invocation of the methods is performed from a Smartphone or a PC, while the execution will be on a server that returns an understandable result through WSDL (Web Service Description Language) generated by web services, where the transport is provided by SOAP (Simple Object Access Protocol).

Internal organs of a human body have very complex structure owing to their anatomic organization. Several image segmentation techniques fail to segment the various organs from medical images due to simple biases. Here, a modified version of the level set method is employed to segment the stomach from CT images. Level set is a model based segmentation method that incorporates a numerical scheme. For the sake of stability of the evolving zero’th level set contour, instead of periodic reinitialization of the signed distance function, a distance regularization term is included. This term is added to the energy optimization function which when solved with gradient flow algorithms, generates a solution with minimum energy and maximum stability. Evolution of the contour is controlled by the edge indicator function. The results show that the algorithm is able to detect inner boundaries in the considered CT stomach images. It appears that it is also possible to extract outer boundaries as well. The results of this approach are reported in this paper.

By Optimal Parity Assignment (OPA) approach for hiding secret data in palette images, we consider the fastest Optimal Parity Assignment (FOPA) methods which are OPA and do not take any extra reordering procedures on color space of palettes. We show that our rho-method is a FOPA and it is easily implemented for controlling quality of stego-images. As consequences, two algorithms for hiding data in palette images are presented. The first algorithm is taken on the rho-forest obtained by rho-method on a palette and the second is only on the colors which are not isolated (by distance not far from the others) in the palette. To prevent from steganalysis, by controlling high quality of stego palette images, combinations of FOPA method with enhanced CPT schemes for binary images are introduced. Some experimental results are presented.

This work shows how to set shape rules and convert them into logical rules to skip incorrect templates and reduce the number of candidate templates in the spatial topology distortion method [1]. The recognition rate is also improved by including shape constraints in the self-organizing process. This will drastically reduce the number of computations with improved recognition.

Due to cost consideration, the quality of images captured from surveillance systems usually is poor. To restore the super-resolution of face images, this paper proposes to use Orthogonal Locality Preserving Projections (OLPP) to preserve the local structure of the face manifold and General Regression Neural Network (GRNN) to bridge the low-resolution and high-resolution faces. In the system, a face is divided into four blocks (forehead, eyes, nose, and mouth). The super-resolution process is applied on each block then combines them into a complete face. Comparing to existing methods, the proposed method has shown an improved and promising result.

Note symbol recognition plays a fundamental role in the process of an OMR system. In this paper, we propose new approaches for recognizing notes by extracting primitives and assembling them into constructed symbols. Firstly, we propose robust algorithms for extracting primitives (stems, noteheads and beams) based on Run-Length Encoding. Secondly, introduce the concept of interaction field to describe the relationship between primitives, and define six hierarchical categories for the structure of notes. Thirdly, propose an effective sequence to assemble the primitives into notes, guided by the mechanism of giving priority to the key structures. To evaluate the performance of those approaches,wepresent experimental results on real-life scores and comparisons with commercial systems. The results show our approaches can recognize notes with high-accuracy and powerful adaptability, especially for the complicated scores with high density of symbols.

The research on network vulnerability analysis and management has gained increased attention during last decade since many studies have proved that combination of exploits is typical means to compromise a network system. This paper presents an intelligent method for analyzing and classifying vulnerabilities based on text mining technology. The proposed mechanism can automatically classify vulnerabilities into different predefined categories and obtain valuable information from abundant vulnerability texts. A series of experiments on 1060 new reported vulnerabilities in last three years by CERT are performed to demonstrate the efficiency of this mechanism. The results generated by this study can be applied to detecting multistage attack, correlating intrusion alerts, and generating attack graph.

The paper describes design of an intelligent information system for assessment of dermatoglyphic indices of Down’s syndrome in infants. The system supports medical diagnosis by automatic processing of dermatoglyphic prints and detecting features indicating presence of genetic disorders. Application of image processing and pattern recognition algorithms in pattern classification of fingerprints and prints of hallucal area of the sole is described. Application of an algorithm based on multi-scale pyramid decomposition of an image is proposed for ridge orientation calculation. A method of singular points detection and calculation of ATD angle of the palm print is presented. Currently achieved results in dermatoglyphic prints enhancement, classification and analysis are discussed. Scheme used in classification of dermatoglyphic prints is described. RBF and triangular kernel types are used in the training of SVM multi-class systems generated with one-vs-one scheme. Results of experiments conducted on the database of Collegium Medicum of the Jagiellonian University in Cracow are presented.

The aim of the article is to show correction possibilities of digital images, achieved by image acquisition tools built on low class CCD matrices, such as their quality were close to images achieved by high class tools. For this purpose the authors used the linear filter with 3x3 mask, which were generated with neural network. Digital images were compared using the quality metrics such as MSE, NMSE and Q.

Container terminals play an important role in marine transportation; they constitute transfer stations to multimodal transport. In this paper, we study the storage of containers. We model the seaport system as a container location model, with an objective function designed to minimize the distance between the vessel berthing locations and the storage zone. Due to the inherent complexity of the problem, we propose a hybrid algorithm based on genetic (GA) and simulated annealing (SA) algorithm. In this paper, three different forms of integration between GA and SA are developed. In order to prove the efficiency of the HGSAAs proposed are compared to the optimal solutions for small-scale problems of an exact method which is Branch and Bound using the commercial software ILOG CPLEX. Computational results on real dimensions taken from the terminal of Normandy, Le Havre port, France, show the good quality of the solutions obtained by the HGSAAs.

The increasing size and complexity of communication satellites has made the manual management of their payloads by engineers through computerised schematics difficult and error prone. This article proposes to optimise payload reconfigurations for current and next generation satellites using a novel Integer Linear Programming model (ILP), which is a variant of network flow models. Experimental results using CPLEX demonstrate the efficiency and scalability of the approach up to realistic satellite payloads sizes and configurations.

In this paper, we propose a global optimization method based on DC (Difference of Convex functions) programming and DCA (DC Algorithm) involving cutting plane techniques for solving two-dimensional Bin packing and Strip packing problems. Given a set of rectangular items, we consider problems of allocating each item to larger rectangular standardized units. In two-dimensional bin packing problem, these units are finite rectangles, and the objective is to pack all the items into the minimum number of units. In two-dimensional strip packing problem, there is a single standardized unit of given width, and the objective is to pack all the items within the minimum height. These problems are characterized as BLP (Binary Linear Programming) problems. Thanks to exact penalty technique in DC Programming, the BLP can be reformulated as polyhedral DC program which can be efficiently solved via the proposed DC programming approach. Computational experiments on large-scale dataset involving up to 200 items show the good performance of our algorithm.

In this paper, a Gaussian Kernel version of the Minimum Sum-of-Squares Clustering

$(G\mathcal{K}MSSC)$

is studied. The problem is formulated as a DC (Difference of Convex functions) program for which a new algorithm based on DC programming and DCA (DC Algorithm) is developed. The related DCA is original and very inexpensive. Numerical simulations show the efficiency of DCA and its superiority with respect to K-mean, a standard method for clustering.

Relational database (RDB) is an adequate tool for rendering concepts, their attributes and relations between them that relate to a target domain. However, RDB has a certain limitation that does not fully allow it to render the semantics or meanings within concepts and their relations even though it has many advantages in representing the relations between concepts. In this paper, we propose a framework which can automatically construct an ontology from an RDB schema. We try to help understand about data structure by clearly identifying the semantic relations between data through the ontology construction. The ontology is constructed on the better understanding about data structure and acts as an assistant tool to efficiently query the data from RDB.

Principal component analysis (PCA) is a well-known method for dimensionality reduction while maintaining most of the variation in data. Although PCA has been applied in many areas successfully, one of its main problems is the sensitivity to noise due to the use of sum-square-error. Several variants of PCA have been proposed to resolve the problem and, among the variants, robust fuzzy PCA (RF-PCA) demonstrated promising results, which uses fuzzy memberships to reduce noise sensitivity. However, there are also problems in RF-PCA and convergence property is one of them. RF-PCA uses two different objective functions to update memberships and principal components, which is the main reason of the lack of convergence property. The difference between two objective functions also slows convergence and deteriorates the solutions of RF-PCA. In this paper, a variant of RF-PCA, called improved robust fuzzy PCA (RF-PCA2), is proposed. RF-PCA2 uses an integrated objective function both for memberships and principal components, which guarantees RF-PCA2 to converge on a local optimum. Furthermore, RF-PCA2 converges faster than RF-PCA and the solutions are more similar to desired ones than those of RF-PCA. Experimental results with artificial data sets also support this.

This paper shows a mathematical foundation for almost important features in the problem of discovering knowledge by association rules. The class of frequent itemsets and the association rule set are partitioned into disjoint classes by two equivalence relations based on closures. Thanks to these partitions, efficient parallel algorithms for mining frequent itemsets and association rules can be obtained. Practically, one can mine frequent itemsets as well as association rules just in the classes that users take care of. Then, we obtain structures of each rule class using corresponding order relations. For a given relation, each rule class splits into two subsets of basic and consequence. The basic one contains minimal rules and the consequence one includes in the rules that can be deducted from those minimal rules. In the rest, we consider association rule mining based on order relation min. The explicit form of minimal rules according to that relation is shown. Due to unique representations of frequent itemsets through their generators and corresponding eliminable itemsets, operators for deducting all remaining rules are also suggested. Experimental results show that mining association rules based on relation min is better than the ones based on relations of minmin and minMax in terms of reduction in mining times as well as number of basic rules.

This paper describes integrity mechanism between OLTPs and offline data. Normally every RDBMS supports five Integrity Constraints (ICs) namely primary key or composite key, unique key, foreign key, not null and check constraints. Online database integrity is achieved through these five ICs. However, as per the retention period data is backed up and removed from the OLTPs for space and performance efficiency. But there is no standardized protocol on keeping integrity between offline data and data present in the OLTPs. Therefore, we present a solution to address the problem of offline data integrity by keeping a representative set of purged data & ICs in the online database to ensure data integrity between OLTPs and offline data. We further support our proposed solution with the help of two types of integrity tests i.e., sufficient and complete test.

This paper describes novel criterion to evaluate QoS of Localization Based Services (LBSs). Architecture of LBSs and communication between particular parts are described. Localization systems based on different technologies are briefly introduced. Parameters required to evaluate QoS of location based services are described. Sufficient range of values for each of the described parameters are introduced and used to evaluate QoS of different location based services based on their demands.

The paper deals with general concept of modular and portable localization system that utilizes existing radio network infrastructure. The modularity means multiple independent collections of algorithms and technologies that allow determining geographical position in various radio and geographical environments. The portability of the system is provided by implementation into small pocket-sized device. The environments are characterized mostly by the parameters of available radio systems such as Global System for Mobile Communications (GSM), Institute of Electrical and Electronics Engineers (IEEE) 802.11 standard-based ones or Global Navigation Satellite Systems (GNSS). Suitable software and utilization of a portable device equipped with necessary hardware can turn the system into the provider of ubiquitous positioning service.

The area of smart workplaces can be considered as one of the most challenging areas for Ambient Intelligence applications. Special focus here can be detected on such typical smart workplaces as smart offices or smart classrooms. In our paper we present some related works specifying more the area of smart workplaces, with accent on their adaptability features. Further on, some approaches to possibilities of user adaptivity concept implementation in smart environments, especially in some types of smart workplaces, will be presented and briefly discussed.

Researchers along the whole world developed many User Adaptive solutions for various devices include mobile devices. The focus is often targeted to one device or group of devices with closer hardware parameters. Developing of universal adaptive GUI Solution is however still unfinished due to developing of new modern devices with many of new parameters and possibilities which is difficult to cover all. Our paper present a solution for “only” one platform – Android, which is however possible to run on various hardware platforms from small mobile Smartphones to large multimedia TV centers. Solution is based on the well-known vector graphics as well as on model view controller (MVC) design pattern. Solution was successfully tested on two user application for Android Market.

This work provides a visualization tool for researchers to explore the geometry of the distribution of human protein-coding DNA in three-dimensional space by applying various manifold learning techniques, which preserve distinct relations among genes. The simulations suggest that the relations hidden among genetic sequences could be explored by the manifolds embedded in Euclidean space. Operating this software, users are able to rotate, scale and shift the three-dimensional spaces in an interactive manner.

Cardiotocography is a biophysical method of fetal state assessment based on analysis of fetal heart rate signal (FHR). The computerized fetal monitoring systems provide a quantitative evaluation of FHR signals, however the effective methods for fetal outcome prediction are still needed. The paper proposes a two-step analysis of fetal heart rate recordings that allows for prediction of the fetal distress. The first step consists in classification of FHR signals with Weighted Fuzzy Scoring System. The fuzzy inference that corresponds to the clinical interpretation of signals based on the FIGO guidelines enables to designate recordings indicating the fetal wellbeing. In the second step, the remained recordings are classified using Lagrangian Support Vector Machines (LSVM). The evaluation of the proposed procedure using data collected with computerized fetal surveillance system confirms its efficacy in predicting the fetal distress.

The results presented in this article introduce the possibility of software processing by image data from CT and MRI in clinical practice. It is important to work with the most accurate data in the diagnosis and further monitoring of the patient. Especially in case of the birth defects or post-traumatic conditions of head called Hydrocephalus, it is necessary to work with this data. A production increase of the cerebrospinal fluid, called cerebrospinal fluid (CSF), causes bring intracranial pressure up. The oppression of the brain tissue has resulted of this procedure. The determination of CSF ratio to the skull in medical practice is used to improve diagnosis and monitoring before and after surgery in patients with Hydrocephalus diagnosed. Software was implemented in Matlab2006b using Image processing Toolbox. Next, the article also describes the design of hardware solutions to these methods of real-time image processing using FPGA programmable logic and genetic algorithm.

The research is focused on the study of the impact of the number and the pattern of geometrical entities and colour models in map like drawings upon the file size of PNG format. The main outputs originate from the exploration of an extended sample set of PNG images generated by Web Map Services of selected European servers. Original images were subsequently transformed to different colour models of PNG (RGBA, RGB, Palette, interlaced, non-interlaced). The images were analysed according to the number of entities, style and size of lines and the number of used RGBA values. We found that the number of geometrical entities can be replaced by the ratio of foreground pixels in an image. The file size grows with higher image density according to different functions which are driven by colour models and line widths, partly also by patterns. In order to minimise files sizes it is recommended to transform images into appropriate palette models and to avoid anti-aliasing changes of transparency.

Modern Smartphones bring to their users many advantages from well-known text prediction with T9 up to voice recognition or personal intelligent voice assistant. These new application features require more and more CPU power which makes a press to CPU producer to develop better and more powered CPUs. New trend can be recognized in multithreading as well as in multicore CPUs. This paper deals with threading possibilities of current versions of mobile devices with Windows Mobile or Android platform where we present a simple architecture to process data in hard or soft realtime with precise time base. We also deal with a future of Smartphones and possible trends to develop multithread application.

In this work, an attempt has been made to analyze retinal images for Content Based Image Retrieval (CBIR) application. Different normal and abnormal images are subjected to vessel detection using Canny based edge detection method with and without preprocessing. Canny segmentation using morphological preprocessing is compared with conventional Canny without preprocessing and contrast stretching based preprocessing method. Essential features are extracted from the segmented images. The similarity matching is carried out between the features obtained from the query image and retinal images stored in the database. The best matched images are ranked and retrieved with appropriate assessment. The results show that it is possible to differentiate the normal and abnormal retinal images using the features derived using Canny with morphological preprocessing. The recall of this CBIR system is found to be 82% using the Canny with morphological preprocessing and is better than the other two methods. It appears that this method is useful to analyze retinal images using CBIR systems.

This paper presents a method for potential topic discovery from blogsphere. A potential topic is defined as an unpopular phrase that has potential to spread through many blogs. To discover potential topics, this method learns from topic frequency transitions in blog articles. Though this learning requires sufficient amount of labeled data, labeled data is costly and time consuming. Therefore this method employs a semi-supervised learning to reduce labeling cost. First, this method extracts candidates of potential topics from categorized blog articles. To detect potential topics from the candidates, a classifier is built from topic frequency transition data. Experimental results with real world data show the effectiveness of the proposed method.

Sensor networks are deployed in numerous military and civil applications, such as remote target detection, weather monitoring, weather forecast, natural resource exploration and disaster management. Despite having many potential applications, wireless sensor networks still face a number of challenges due to their particular characteristics that other wireless networks, like cellular networks or mobile ad hoc networks do not have. The most difficult challenge of the design of wireless sensor networks is the limited energy resource of the battery of the sensors. This limited resource restricts the operational time that wireless sensor networks can function in their applications. Routing protocols play a major part in the energy efficiency of wireless sensor networks because data communication dissipates most of the energy resource of the networks. This paper studies the importance of considering neighboring nodes in the energy efficiency routing problem. After showing that the routing problem that considers the remaining energy of all sensor nodes is NP-complete, heuristics are proposed for the problem. Simulation results show that the routing algorithm that considers the remaining energy of all sensor nodes improves the system lifetime significantly compared to that of minimum transmission energy algorithms.

Agent-based ubiquitous computing (AUC) is a form of distributed computing by which computational processes are executed concurrently by assigning each computational process to one of agents on a ubiquitous computing system (UCS). One of AUC goals is to support the seamless integration of information and services. Meeting this grand challenge of AUC requires that agent-orientation not tackled before is necessarily featured. To this end, this paper presents a firm formal development for featuring agent-orientation of ubiquitous computing to integrate smoothly information and services.

In this paper, a new approach using an Modular Radial Basis Function Neural Network (M-RBF-NN) technique is presented to improve rainfall forecasting performance coupled with appropriate data–preprocessing techniques by Singular Spectrum Analysis (SSA) and Partial Least Square (PLS) regression. In the process of modular modeling, SSA is applied for the time series extraction of complex trends and finding structure. In the second stage, the data set is divided into different training sets by used Bagging and Boosting technology. In the third stage, then modular RBF–NN predictors are produced by different kernel function. In the fourth stage, PLS technology is used to choose the appropriate number of neural network ensemble members. In the final stage, least squares support vector regression is used for ensemble of the M-RBF-NN to prediction purpose. The developed RBF-NN model is being applied for real time rainfall forecasting and flood management in Liuzhou, Guangxi. Aimed at providing forecasts in a near real time schedule, different network types were tested with the same input information. Additionally, forecasts by M-RBF-NN model were compared to the convenient approach. Results show that that the predictions using the proposed approach are consistently better than those obtained using the other methods presented in this study in terms of the same measurements. Sensitivity analysis indicated that the proposed M-RBF-NN technique provides a promising alternative to rainfall prediction.

Given a connected, weighted, undirected graph G = (V, E), a set infrastructure nodes and a set customers C includes two customer types where by customers C1 require a single connection (type-1) and customers C2 need to be redundantly connected (type-2). Survivable Network Design Problem (SNDP) seeks sub-graph of G with smallest weight in which all customers are connected to infrastructure nodes. This problem is NP-hard and has application in the design of the last mile of the real-world communication networks. This paper proposes a new heuristic algorithm for solving SNDP. Results of computational experiments are reported to show the efficiency of proposed algorithm.

Indirect immunofluorescence (IIF) with HEp-2 cells has been used to detect antinuclear auto-antibodies (ANA) for diagnosing systemic autoimmune diseases. The aim of this study is to develop an automatic scheme to identify the fluorescence patterns of HEp-2 cell in IIF images. The self-organizing map (SOM) neural network with 14 textural and statistic features were utilized to classify the fluorescence patterns. This study evaluated 1020 autoantibody fluorescence patterns that were divided into six pattern categories, i.e. diffuse, peripheral, coarse speckled, fine speckled, discrete speckled and nucleolar patterns. Experimental results show that the proposed approach can identify autoantibody fluorescence patterns with a high accuracy and is therefore clinically useful to provide a second opinion for diagnosing systemic autoimmune diseases.

In spite of the gradual increase of the academic studies on smuggling crime, they seldom focus on the subject of applying data mining to crime prevention. Artificial Neural Networks and Logistic Regression are used to conduct classification and prediction. This study establishes models for vessels of different tonnage and operation purpose, which can provide the enforcers with clearer judgment criteria. The study results show that the application of Artificial Neural Networks to smuggling fishing vessel can get the average precision as high as 76.49%, the application of Logistic Regression to smuggling fishing vessel can get the average precision as high as 61.58%, both of which are of significantly higher efficiency compared with human inspection. The information technology can greatly help to increase the probabilities of seizing smuggling vessels, what’s more, it can make better use of the data in the database to increase the probabilities of seizing smuggling crimes.

A millions of data has been produce by cross-organizational research and collaborations must be managed, shared and analyzed. Data grid is a useful technique to solve these tasks that applicable to process the large number of data produced by scientific experiments. It also enables an organization to operate and manage distributed resources over the internet as a secure, robust, and flexible infrastructure. Some problems must be considered in managing data grid such as reliability and availability of the data to the user access, network latency, failures or malicious attacks during execution and etc. The replication strategy is the solution to solve these problems that can minimize the time access to the data by creating many replicas and storing replicas in appropriate locations. In this paper, we present some reviews on the existing dynamic replication replacement strategies due to the limited storage used on data grid and also to improve the management of data grid. It is shown that replication techniques able to improve availability and reliability of data, network latency, bandwidth consumption, fault tolerance and etc in data grid environments.

The cloud is a next generation platform that provides dynamic resource pools, virtualization, and high availability. The concept of cloud computing is using a virtual centralization. This means, in one part, we have a full control on data and processes in his computer. On the other part, we have the cloud computing where the service and data maintenance is provided by vendors. The client or customers usually unaware about the place where processes are running or the data is stored. So, logically speaking, the client has no control over it. This is the reason cloud computing facing so many security challenge. In this paper, we presented selection issues in cloud computing and focus on the security issues. There are four cloud computing security issues that will be focused, namely XML signature, browser security, cloud integrity and binding issues and flooding attacks. Data security on the cloud side is not only focused on the process of data transmission, but also the system security and data protection for those data stored on the storages of the cloud side. There are some considerations that need to be focused in order to achieve better safe environment in cloud computing such as storage and system protection and data protection. In order to achieve better performance in security, cloud computing needs to fulfill five goals which are availability, confidentiality, data integrity, control and audit. By implementing these goals, we hope data security in cloud computing will be more secure. We also hope that cloud computing will have a bright future with arise of a large number of enterprises and will bring an enormous change in the Internet since it is a low-cost supercomputing to provide services.