One of the most advantages of the Semantic Web is to augment the data with a well-defined meaning and linking between data by using the OWL ontology language. Today most of data are stored in relational databases. In order to reuse and infer this data on the Semantic Web, there is a need for converting the data stored in relational databases to the form of OWL. Some approaches have been proposed, however, most of them transform a single table into OWL individuals. This paper presents RDB2OWL, a complete method to transform all tables in the relational database into OWL ontology. The transformation makes it possible to reverse OWL ontology to relational tables. Most of all, all the steps in RDB2OWL are done automatically without any user intervention.

This paper presents a social network with a peer-to-peer architecture that facilitates social computing services in distributed environment. This social network aims at providing users the capability of managing the dissemination of user data, searching user data on the data silos of the network, and consolidating user data from various social networks. The social network employs a super peer peer-to-peer architecture that contains peers and super peers. Users use peers to participate the network and services. Peers with sufficient storage, bandwidth and processing power become super peers that support peers for complex operations such as user authentication or group communication. We have extended the Gnutella protocol to provide the authentication and posting services on the social network. The design of these services copes with the distributed setting of the social network. The evaluation of the prototyping social network has performed on a number of laboratory workstations to investigate its scalability and reliability.

This research explores the mediating role of the attitude toward using e-learning in the association between the perception on the usefulness of e-learning and the adoption of e-learning as well as the moderating role of the perception on the usefulness of e-learning in the relationship between the attitude toward using e-learning and the adoption of e-learning. We use Sobel's analysis and hierarchical regression analysis to examine the mediating and moderating relationships in our research. The findings reveal that the attitude toward using e-learning intervenes in the association between the perception on the usefulness of e-learning and the adoption of e-learning. The perception on the usefulness of e-learning is statistically evidenced to moderate the association between the attitude toward using e-learning and the adoption of e-learning. This research will be significant to education institutions in their decision to adopt e-learning for their education program.

Error localization is a critical issue in software debugging. When a long and complex program contains errors, it is difficult even for experienced programmers to localize the portions of the program that contain errors. In this paper, we present a new approach to assist programmers in localizing errors in programs. We convert C programs into constraint satisfaction problems, and localize errors by finding inconsistent constraints. In addition, we develop a tool for localizing errors automatically. Extensive numerical experiments indicate that the constraint based approach is very promising.

Web services transactions can be dened as a sequence of messages exchanges between services to achieve a mutually agreed outcome. This paper proposes an approach to verify the reliability of Web services transactions. In this approach, the reliability of Web services transactions are specied by temporal operators, especially using past temporal operators in temporal logic. Web services models are specied by nite-state automata. We use NuSMV model checker to verify if the Web services model satises their transaction properties. We use well known transactions of the Virtual Travel Agency Web services to illustrate our approach.

Low error rate and high capacity multimedia delivery are urgently required for next generations of wireless communication systems with multiple hops and multiple antennas. Asymptotic capacity and precoding designs have been presented in the literature for some cases of wireless correlated multi-hop multi-input multi-output (MIMO) networks. However, an optimal precoding design for bit-error-rate (BER) performance improvement has not been well investigated. In this paper, we design the precoding matrix for the general case of correlated wireless multi-hop MIMO channels. The channel at each hop is spatially correlated, the source symbols are mutually correlated, and the additive Gaussian noise at each hop is colored. Since the optimal precoding design to minimize the error rate cannot be analytically obtained in closed-from as the optimization problem is very complicated and neither convex nor concave, we propose to relax the optimization problem by considering the design problem individually to obtain a sub-optimal design in closed-form. Simulation results show that the proposed precoding design can significantly reduce BER and increase end-to-end mutual information while it does not require resources of the system such as transmission power or bandwidth.

This paper investigates the impact of equally correlated Rayleigh fading on the performance of underlay cognitive single-input multiple-output (SIMO) systems with selection combining. In particular, the exact closed-form expression of the system outage probability (OP) is derived over Rayleigh fading channels. To gain more insights, the asymptotic expression of OP is provided. It is shown that the system diversity is a function of the correlation coefficient. Monte-Carlo simulations confirm the correctness of the suggested analysis approach.

In this paper, we analyze the secrecy performance of the relay selection methods in interference-limited networks. In particular, we propose the optimal and suboptimal relay selection methods to enhance the secrecy rate. For performance evaluation, we derive exact and asymptotic closed-form expressions of secrecy outage probability over Rayleigh fading channel. Monte Carlo simulations are performed to validate our derivations.

In this paper, a full-diversity high-rate space time block code (STBC) is proposed for MIMO system with two transmit antennas. The proposed STBC achieves full diversity with M-QAM modulation by constellation stretching. Simulation results show that the proposed STBC outperforms significantly previous code which has the same code rate, especially for high order modulations.

A capacity analysis for the general case of spatially correlated wireless multi-hop multi-input multi-output (MIMO) networks is presented in this paper. First, we derive the optimal source symbol covariance for the maximum mutual information between the channel input and the channel output when having the full knowledge of the channel at the transmitter. Secondly, we formulate the average mutual information maximization problem when having only the channel statistics at the transmitter. Since this problem is impossible to be solved analytically, the numerical interior-point-method is employed to obtain the optimal solution. Furthermore, to reduce the computational complexity, an asymptotic closed-form solution is derived by maximizing an upper bound of the objective function. Simulation results show that the average mutual information obtained by the asymptotic design is very closed to that obtained by the optimal design, while saving a huge computational complexity.

In this paper, a new ultra-wide band (UWB) MIMO antennas for UWB application are proposed. MIMO antennas consists of two single ultra-wide band antenna. This MIMO antennas are designed with the working bandwidth from 3.1 GHz to 10.6 GHz. The purpose is to design a simple and compact antennas which has a broad bandwidth VSWR ≤ 2. MIMO Antennas characteristic such as radiation pattern, maximal gain are also thoroughly investigated.

This paper presents a method to optimize structure of planar inverted F antenna by meandering and folding technique for the monopole antenna placed on FR4 dielectric. The designed antenna has compact size 21 × 14.5 × 4 mm3. Moreover, this antenna still offer enough wide bandwidth > 270 MHz (VSWR ≤ 2), which covers 3G band (VSWR ≤ 2). Using the simulation program to optimize antenna structure and calculate the antenna parameters in order to verify its applicability for the 3G devices.

This paper presents a 915MHz ultra-low power receiver design. The receiver is a RF envelope detection receiver with OOK modulation scheme. It is designed using 65nm CMOS technology. Simulation results show that the receiver consumes only 120uW at 0.5V supply voltage while its sensitivity can reach -86dBm with a data rate of 500Kbps.

In an information-centric network, data objects have a central role. They are located solely based on their name, and accessed regardless of their location. The new paradigm helps overcome the limitations of the host-centric model, thereby offering a wide range of applications. However, as it is at an early stage, the Information-Centric Networking concept solicits decent techniques and architectures. In this paper, we propose an architectural design for Information-Centric Networking. Our work is expected to contribute towards the overall development of the research area.

This paper presents the design of a wideband circularly polarized E-shaped patch antenna for 2.4-GHz wireless local area networks (WLAN) applications. The proposed antenna is a modified form of the conventional circularly polarized E-shaped patch antenna. By incorporating additional slots into the antenna patch, the impedance bandwidth and return loss of the circularly polarized antenna are improved by about 6.5% and 12 dB, respectively. Measurements of the fabricated antennas show good agreement with simulated results.

Segmenting reddish lesions in capsule endoscopy (CE) images is an initial step for further computer-assisted application such as image enhancement, abnormal measurement/tracking, and so on. In this paper, we propose an automatic segmentation method that is successful even with CE image including unclear reddish lesions. To obtain this, the proposed scheme seeks good features to discriminate the reddish lesions from normal tissues. For implementations, we first extract only meaningful regions in a CE image. Using the proposed features, the meaningful regions on CE image is extracted. We consider a corresponding local mean image at a certain scale. Candidates of the abnormal regions are located in the local mean image with assistants of a diffusion process. Because the major issue involves finding out an optimal scale, we utilize an iteration procedure to select such value. Evaluations in the experiments confirm effectiveness of the proposed method with both qualitative and quantitative measurement.

This paper describes an efficient combination KINECT data (RGB and Depth data) for background subtraction. To obtain this goal, we simply utilize a statistic model of background pixels like Gaussian Mixture Model for color and depth features. However, beyond results of the segmentation from separated data, our combination strategy takes into account spatial pixels whose depth (or color feature) is more valuable than the other one. Our strategy is that in valid range of the depth measurement, results of foreground segmentation from depth features are biased, whereas the pixels where out of the range, results of foreground segmentation using color feature are utilized. Following such combination scheme, depth pixels are filtered through a proposed noise model of depth as well as is validated in a range of the depth measurements. The proposed method is evaluated using a public dataset which is suffered from common problems of the background subtraction such as shadows, reflections and camouflage. The experiments show segmentation results which are comparable with recent reports. Furthermore, the proposed method is successful with a challenging task such as extracting human fall-down in a RGB-D image sequence. The foreground segmentation results is feasibility for recognition task.

Recently, abnormal event detection has attracted great research attention because of its wide range of applications such as elderly surveillance. In this paper, we propose an hybrid method combining both tracking output and motion templates. This method consists of two steps: object detection, localization and tracking and abnormal event detection. Our contributions in this paper are three-folds. Firstly, we propose a method that apply only HOG-SVM detector on extended regions detected by background subtraction. This method takes advantages of the background subtraction method (fast computation) and the HOG-SVM detector (reliable detection). Secondly, we do multiple objects tracking based on HOG descriptor. The HOG descriptor, computed in the detection phase, will be used in the phase of observation and track association. This descriptor is more robust than usual grayscale (color) histogram based descriptor. Finally, we propose a hybrid method for abnormal event detection this allows to remove several false detection cases.

Most studies on sequence data mining have focused on mining all possible frequent sequences. However, this produces redundant results, increasing required the storage space and run time, especially for large sequence databases. Mining frequent closed sequences can fully extract necessary information. Some algorithms for mining frequent closed sequences have been proposed with most using a candidate maintenance-and-test paradigm. The present paper proposes an algorithm, CloFS-DBV, that uses dynamic bit vectors. Various methods are employed to reduce memory usage and run time. Experimental results show that CloFS-DBV is more efficient than the BIDE algorithm in terms of execution time and memory usage.

This paper proposes a framework for clustering data streams, which we call SYNSTREAM (SYNchronization-based STREAM Clustering), by using the synchronization of pulse-coupled oscillator network. SYNSTREAM is an unsupervised clustering method in which the optimum number of clusters can be controlled. SYNSTREAM can build clustering structure from data stream in self-organized manner through the self-synchronization process. SYNSTREAM does not involve optimizing an objective function. Our empirical experiments show that our algorithm can be used for summarizing the sensor data streams eciently.

The low level of received signals power makes Global Navigation Satellite Systems (GNSS) receivers vulnerable to many classes of disturbing signals. Among them, narrow band interference (NBI) might cause serious receiver performance degradation. Cancellation of NBI can be implemented by using notch filters (NF), which are controlled by two parameters: the notch frequency, which specifies the band center, and the notch bandwidth, which defines the spectrum area to be removed. Many literatures on the topic are focused on adapting the notch frequency, without estimating the filter bandwidth. This paper proposes a method able to determine both the notch parameters, optimizing the interference suppression.

This paper introduces a traffic engineering routing algorithm that aims to accept as many routing demands as possible on the condition that a certain amount of bandwidth resource is reserved for each accepted demand. The novel idea is to select routes based on not only network states but also information derived from routing data such as probabilities of the ingress egress pairs and usage frequencies of the links. Experiments with respect to acceptance ratio and computation time have been conducted against various test sets. Results indicate that the proposed algorithm has better performance than the existing popular algorithms including Minimum Interference Routing Algorithm (MIRA) and Random Race based Algorithm for Traffic Engineering (RRATE).

This paper presents a novel router architecture for implementing a Reconfigurable Network-on-Chip (RNoC) at high level design using SystemC language. RNoC is an adaptive NoC-based system-on-chip providing a dynamic reconfigurable communication mechanism. By adding a virtual port -- named Routing Modification port -- into the router architecture, the network router is able to route communication data flexibly whenever the target routing path is blocked, by unwanted defects or intently by the programmation to meet the requirements of applications. The proposed architecture has been modeled in SystemC, simulated and verified within a 2D mesh 5x5 network platform. The static XY routing algorithm has been used in the normal communication mode while the West-First algorithm with a proposed prohibited router surrounding technique has been applied in the reconfiguration mode. Experimental results are also reported to compare the performance of the network architecture in different operation modes.

In this paper, the node-centric formulation for the inelastic traffic in multi-hop wireless network is considered. The network utility maximization (NUM) is a nonconvex optimization problem due to the nonconcavity of the sigmoidal functions representing the inelastic flows. NUM cannot be solved by the canonical methods. We approximate its equivalent problem to a convex one which can be solved by the dual-based decomposition approach. By successively solving the approximations, the stationary point of the jointly rate control, routing, and scheduling algorithm converges to a KKT solution of the original problem.

This paper is concerned with the problem of multiuser trans- mission in a multiple input multiple output (MIMO) two-way relay net- work. In the network with multiple multi-antenna users, multiuser pre- coding and power allocation is developed to maximize the network sum- rate of both uplink and downlink under power constraints at nodes. To justify the performance of the proposed precoding and power allocation technique, simulated network sum-rate values are provided under vari- ous system settings for comparison between the proposed technique and other existing ones.

Very recently in the work "Simple Iterative Method for Solving Problems for Plates with Partial Internal Supports, Journal of Engineering Mathematics, DOI: 10.1007/s10665-013-9652-7 (in press)", we proposed a numerical method for solving some problems of plates on one and two line partial internal supports. In the essence they are problems with strongly mixed boundary conditions for biharmonic equation. Therefore, the method combines the reduction of the order of the equation with a domain decomposition technique. In this paper we apply the method to plates on internal supports of more complicated configurations. Namely, we consider the case of three line partial internal supports, where the computational domain of the problem is divided into four subdomains and the cases of cross support, rectangular support and H-shape support, where the computational domain of the problems is divided into three subdomains . The convergence of the method is established theoretically as in the mentioned above paper for one line internal support and especially, its efficiency is confirmed on numerical experiments.

One-class Support Vector Machine (OCSVM) is a well-known kernel-based method for one-class classification problem. Although OCSVM renders the good performance for the imbalanced data set, its two obvious drawbacks are that OCSVM cannot employ the negative data samples and its performance may be downgraded by the noisy and corrupted data. In this paper, we first extend the model of OCSVM for enabling it in using the negative data samples for classification and propose how to combine it with the fuzzy framework for handling the noisy and corrupted data.

A symmetric necklace is a necklace that this necklace and its reverse are in the same equivalence class under rotation. An efficient algorithm is the one that uses only a constant amount of computation per object, in an amortized sense. Such algorithms are also said to be constant amortized time (CAT) algorithms. In this paper, we first present some characteristics of symmetric necklaces, then propose an efficient algorithm for generating them and finally count the cardinality of this object class.

Our investigation aims at constructing random oblique decision trees to recognize handwritten digits. At the pre-processing step, we propose to use the GIST descriptors to represent digit images in large number of dimensions. And then we propose a multi-class version of random oblique decision trees based on the linear discriminant analysis that is suited for classifying high dimensional datasets. The experimental results on USPS, MNIST datasets show that our proposal has very high accuracy compared to state-of-the-art algorithms.

This paper addresses the problem of segmenting common objects from a pair of images. We introduce an approach based on branch-and-bound algorithm to find pairs of regions having high similarity scores, given a region pair is formulated by two segmented regions in two images. Experiments conducted on a database of real-world images show that our approach is 3-7 times more efficient than exhaustive search in finding the optimal region pairs.

Evaluating semantic similarity between concepts is a very common component in many applications dealing with textual data such as information extraction, information retrieval, natural language processing, or knowledge acquisition. This paper presents an approach to assess semantic similarity between Vietnamese concepts using Vietnamese Wikipedia. Firstly, the Vietnamese Wikipedia' structure is exploited to derive a Vietnamese ontology. Next, we employ measures of semantic similarity in literature based on the Vietnamese ontology. Then we conduct an experiment providing 30 Vietnamese concept pairs to 18 human subjects to assess similarity of these concept pairs. Finally, we estimate the correlation between human judgments and the results of similarity measures employed using Pearson product-moment correlation coefficient. The experiment results show that our system achieved quite high performance and that similarity measures between Vietnamese concepts are potential in enhancing the performance of applications dealing with textual data.

Ground penetrating radar systems during the measurement data are always affected by the external components. The impact of these components affects the measurement results, deviation of the target component. Measurement process is established, the wavelength of the electromagnetic field generated in the ground, the obstacles encountered target, wave intensity signals were collected, in addition to the components of the frequency data signals always contain frequency noise component. Dewow processing methods are put to the test in order to simultaneously remove low frequency noise affects the data collected.

Currently, the usual method for sleep stage classification is visual inspection by sleep specialist, which is very difficult and time consuming. In this research, we investigated on developing an algorithm for automatic detection of Slow Wave activities as the present of Slow Wave Sleep (SWS) or stage N3, as well as for the evaluation of the deep sleep characteristics based on EEG signal responding to different types of sleep disorders. Polysomnography recordings from twenty seven subjects were used in this study. These subjects were belonging to four groups of sleep diseases - Insomnia, Obstructive Sleep Apnea (OSA), Restless Leg Syndrome (RLS), Snoring and Normal. The definition of sleep stages and the sleep literature have shown that the amplitude of Power Spectra of four EEG frequency components including Alpha wave (8-13Hz), Slow Wave Activity (SWA) or also referred as Delta wave (0.5-2Hz), Theta wave (4-7Hz) and Sigma wave (13-16Hz) vary corresponding with different sleep stages. Based on this, we applied Threshold-based classification on Average Power of these four EEG frequency components to detect stage N3. The Average Power of each frequency band was taken by the trapezoid approximation of the area under the Power Spectra curve in the corresponding frequency area. The results demonstrated that Slow Wave Sleep could be automatically detected in all cases of sleep diseases with specificity from 76.95  1.310% to 88.82  0.583 %, sensitivity from 61.45  1.333% to 82.63  0.368% and an averaged accuracy of 86.83  1.641%. This gave an idea for further research in sleep scoring, sleep analysis and sleep disorder studies about SWS.

In this paper, we propose a novel UHF lowpass filter for stepped frequency continuous wave (SFCW) ground penetrating radar (GPR) system receivers. The novel lowpass filter makes high performance for miniaturization and harmonic suppression of SFCW GPR system receivers. The structure of the lowpass filter is T-shaped transmission line with the developed stub line. The microstrip connection lines between circuits of T-shaped are designed by meander microstrip lines. This design helps to reduce the loss when the lowpass filter is fabricated and used in actual environments. The filter is fabricated in FR4 substrate and copper patch. A good agreement has been observed between the experimental results and simulation data. The filter operates in 0 - 800 MHz and ultra wide-stopband from 1.35 - 4 GHz band.

In this report, we evaluate two ways of combining two popular methods used in detecting Rapid Eye Movement (REM) event that base on Negative Instantaneous Products (NIP) method and slope Control Points (CP) method. First, we describe the characteristic of the source data in comparison with Alice5 system and present two combinations with the improved algorithms. We then report the REM event and its density of 22 patients suffering from five Sleep Disorders (Normal, Insomnia, Snoring, Restless Legs Symptom (RLS), Obstructive Sleep Apnea (OSA) and Narcolepsy). Results are presented along with our validation of the advantages and disadvantages using the analysis and display system. Our research shows a strong correlation between the two combinations, and further comparison with Alice5 from moderate to strong correlation has resulted in significant enhancement of mixing wave periods. The great variation in the density range in slope Control Points (CP) method also promises further application in REM sleep of behavior disorders.

In this paper, we present an experiment to measure the Power Delay Profile (PDP) of Underwater Acoustic Communication (UAC) channel. Transmitter and Receiver are fixed point to point communication. A probing pulse is transmitted successively at the transmitter. At the receiver, received signals are processed through 4 steps. PDP of the channel will be found out by mean of executing various measurements. Some notable parameters of the channel, which affect communication systems powerfully, are analyzed and evaluated. Measurement and analysis results show that the delay spread decreases as the transmission distance increases.

Based on the combination of expert knowledge in aquaculture, the theory of Bayesian networks, and Markov chains, models for forecasting white spot disease (WSD) on black tiger prawns are proposed. These models are built by basing on the pathology, environmental factors, and biological characteristics of black tiger prawns. Models aim to assess the infectious status of WSD in a pond at a specific time, the infectious status of WSD in a pond at adjacent time steps in future, and the possibility of the infection of WSD to neighboring ponds. Using the obtained results, the warning about WSD can be informed soon to farmers.

This paper presents a method for generating the new meaning-summarizing sentences for some Vietnamese paragraph forms which their meanings are represented in Discourse Representation Structures (DRS). Each new generated meaning-summarizing sentence will summarize the meaning of one paragraph. However in the first steps of research, our method is limited for Vietnamese paragraphs composed by two simple sentences related by inter-sentential anaphoric pronouns. Our method is performed through three phases: the first phase is analyzing and identifying the main predicates in DRS structure; the second phase is generating the syntactic structure for the new meaning-summarizing sentence with the algorithm suitable for each paragraph form; the third phase is completing the new meaning-summarizing sentence. The Vietnamese paragraphs which are considered in this research all have the characteristics: the first sentence has one transitive verb having the relation with two nouns indicating person, and the second sentence has one or two pronouns indicating person.

Active learning allows semi-supervised clustering algorithms to solicit domain experts to retrieve a few set of class labels (or seeds) to improve their efficiency or the relevance of their results. However, some recent studies show that, even in the case of a good answer from the domain expert, semi-supervised clustering can see their performances drop with badly chosen seeds. Until now, only few works address the problem of determining the best queries for a clustering algorithm in an active learning context, and most of these studies are limited because of their hypothesis on the size and the shape of expected clusters. In this paper, we propose a new active seed selection algorithm that makes no hypothesis on the underlying data distribution. Experiments conducted on real data sets show the efficiency of this new approach compared to existing ones.

This paper shows an efficient framework for ElectroCardioGraph (ECG) signal processing in a real multi-lead system, including three main functions: de-noising, removing base-line drift, extracting features. This research exploits the potential of Pan-Tompkins algorithm in real-time processing and improves it to be integrated efficiently in a multi-lead ECG system. The proposed framework selected a better filter for de-noising ECG signals and used Q-line compensation to remove base-line drift. Feature extraction processing is optimized in the proposed framework by using the key lead, one of the two leads (lead I and lead II) is to produce the best value for the ECG feature extraction. The experiments show that its results in extracting features of ECG signals have the high accuracy compared with the recent studies. The significant key of the proposed framework is the low complexity and capacity for real-time processing

One-class Support Vector Machine (OCSVM) is a well-known kernel-based method for one-class classication problem. Although OCSVM renders the good performance for the imbalanced data set, its two obvious drawbacks are that OCSVM cannot employ the negative data samples and cannot utilize the unlabeled data to boost the classier. In this paper, we rst extend the model of OCSVM for enabling it in using the negative data samples for classication and propose how to integrate the semi-supervised paradigm to the extended OCSVM for utilizing the unlabeled data to increase the classifier's generalization ability.

Gene regulation, despite being investigated in a large number of works, is still yet to be well understood. The mechanisms that control gene expression is one of the open problems. Epigenetic factors, among others, are assumed to have a role to play. In this work, we focus on DNA methylation and posttranslational histone modifications (PTMs). These individually have been shown to be attribute to controlling of gene expression. However, neither can totally account for the expression level, i.e. low or high. Therefore, the hypothesis of their combinatorial role, as two of the most influencing factors, has been established and discussed in literature. Taking a computational approach based on rule induction, we identified some key PTMs that have considerable effects such as H2BK5ac, H3K79me1/2/3, H4K91ac, and H3K4me3. Also, some interesting patterns of DNA methylation and PTMs that can explain the low expression of genes in CD4+ T cell. The results include previously reported ones as well as some new valid patterns which could give some new insights to the process in question.

In this paper, we investigate two different selection schemes for choosing parental solutions in the direction based multi-objective evolutionary algorithm-II (DMEA-II). With the original selection scheme, parental solutions (which will be used for producing offspring solutions) are randomly selected from the current population. For this investigation, a new selection scheme is also introduced in which parental solutions are assigned a fixed index for selection. In order to analyze the effect of different selection schemes on DMEA-II, we carried out a case study on several well-known test problems with two primary performance metrics, namely the generation distance (GD), and inverse generation distance (IGD). Our analysis on the effective results indicates the better performance of the new selection scheme on DMEA-II.

miRNAs are a class of small non-coding regulatory RNAs that play important roles in the post-transcriptional regulation. Many studies have shown the role of miRNA on human disease. Several network-based methods have been proposed to predict miRNA-disease associations. In those networks, functional similarity interactions based on shared target genes of miRNAs were primarily used. However, recently, functional synergistic interactions between miRNAs have been computationally and experimentally identified. In this study, we combine these synergistic interactions with the functional similarity interactions to improve the performance of disease miRNA prediction. To this end, we constructed functional similarity networks, in which two miRNAs are functionally interacted if they share at least one target gene. Next, we integrate these networks with a functional synergistic network. A random walk with restart algorithm was applied on each of the individual networks and the integrated one to rank candidate miRNAs based on their similarity to known disease miRNAs. A leave-one-out cross-validation method was used to assess the performance of the networks. As a result, the performance was much better in the integrated network than the individual ones measured by the area under the ROC curve (AUC). This also indicated that our method outperformed RWRMDA, a recent state-of-the art method, which used a same RWR-based method solely on a functional similarity network of miRNAs. We additionally tested our method with breast cancer and selected top 100 highly ranked candidate miRNAs. A total of 17 of them were already previously found evidenced to be associated with breast cancer.

The problem of optimizing a real function over the efficient set of a multiple objective programming problem arises in a variety of applications. Because of its interesting mathematical aspects as well as its wide range of applications, this problem has attracted the attention of many authors. In this article, we propose a branch and bound algorithm in outcome space for minimizing a function $h(x)=\varphi(f(x))$ over the efficient set $X{E}$ of the bi-criteria convex programming problem ${\rm Vmin}{f(x)=(f{1}(x),f_{2}(x))^{T}|x\in X}$, where the function $\varphi$ is a quasi-concave continuous and increasing on $f(X)$. The convergence of the algorithm is established. Preliminary computational results with the proposed algorithm are reported.

The adjoint method is effectively applied for identifying a location of the pollutant source in water or air pollution problems. Using the relation between solutions of main (forward) and adjoint (backward) problems, the estimation of pollutant source parameters is defined quite simply. In the paper, we shall utilize the adjoint method for identifying a location source and pollutant power of the stationary atmospheric pollution point source problem. The proposed method is so simple that the pollutant concentration measuring at three points are enough to define unknown parameters. Some numerical examples are presented for illustrating the effectiveness of the method.

In most languages, the quality of a text to speech system is relating directly to the diversity of language' domain. Each domain, such as sports, entertainments, etc… has its own grammar structure that will decide the auto pronunciation of the text to speech system. The prosodic information plays as a crucial role for analyzing the grammar structure of each domain. In this research, we will analyze characteristics of prosodic information of aviation domains in Vietnamese, which are represented by a set of common airline announcements.

This paper proposes a novel approach that improves bag of features (BOF) by using k-means clustering based on coefficient of variation (CV-k-means), called CV-BOF. In the context, our approach is the fusion of CV-BOF and spatial pyramid histogram to solve the loss of spatial information accompanying the features. Our experimental results on the JAFFE, YALE and FERET databases demonstrate that Coefficient of Variation Bag of features can achieve very high recognition accuracy despite of using smaller visual words space. These results indicate that our approach is suitable for building efficient and feasible descriptors to improve the recognition performance.

Random forests have been increasingly used as a statistical method for classification and regression. Selecting good subsets of features is important in learning an accurate Random Forest, especially in high dimensional data. This paper proposes a new technique for feature subspace selection in RF using two-step quality-based feature sampling method. In the first step, we used random forests permutation method to produce raw importance feature scores and then applied statistical method to remove all redundant features or noise. The final step was to filter the features which are highly related to the response feature from less important ones using a correlation function. These steps partition the features of the input data(without noise) into strong and weak important feature subsets. The two-step quality-based feature sampling method independently samples features from the two subsets and puts them together as the subspace features for splitting the data at any node, recursively. This approach enables to generate trees from bagged sample data with lower errors and higher test accuracy. Furthermore, its purposes include reducing dimensionality, reducing the amount of data needed for learning the random forests model. Our experimental results have shown that random forests with this improvement outperformed existing random forests and quantile regression forests in reduction of mean square errors and increasing of prediction accuracy.

A low complexity fusion method for increasing the spatial resolution of VnRedSat-1's multispectral images is proposed in this paper. In this method, a content-based registration process between multispectral and panchromatic images is applied before a simplified fusion. Our proposed method is evaluated with other four well-known fusion methods in both subjective and objective tests. The experimental results exhibit a better image quality and reduced computation time.

In this paper, a computing procedure for stock value and nancial index forecasting based on fuzzy clustering and fuzzy inference system is presented. Firstly, we present a data processing method based on percentage variation rate. Then we construct a fuzzy inference system with fuzzy rules obtained by the fuzzy clustering process. We determine weight of each rule and construct a defuzzication method. Finally, we apply the proposed computing procedure to some nancial forecasting problem such as Vietnam's stock value and foreign exchange. The exper- imental results show that our computing procedure gives better forecasting results in some case than several conventional models such as AR,ANFIS.

This paper represents a semi-autonomous control algorithm combining between an autonomous mode using a stereo camera system and user's intention using an accelerometer sensor in an electrical wheelchair for severely disabled people. In the autonomous mode, the wheelchair will detect obstacles and freespaces for collision avoidance. For the safety of user, the wheelchair autonomously avoids obstacles, if a real distance between the wheelchair and obstacles is too closed. User's intention is that an accelerometer sensor is mounted on the top of a cap for creating control commands. The user with this cap can drive the wheelchair to reach the desired target by moving the head. In the semi-autonomous control strategy, a safe distance is installed in the wheelchair controller so that the wheelchair can autonomously avoid obstacles during its movement. The experimental results with the proposed controller show to illustrate that the wheelchair can move affectively in the indoor environment

In this paper, we present a solution to the Simultaneous Localization and Mapping (SLAM) problem for an indoor robot using bearing-only observations. An omnidirectional camera is used to observe indoor scene from which vertical lines are extracted to obtain bearing measurements. To track vertical lines through sequence of omnidirectional images, a matching algorithm based on histogram of oriented gradients technique is proposed. The Extended Kalman Filter (EKF) is used to estimate the 3-DoF motion of the robot along with two-dimensional positions of vertical lines in the environment. In order to overcome bearing-only initialization, the Unscented Transform is used to estimate the probability distribution function (PDF) of an initialized vertical line. Simulations and real experiments have been carried out to validate the proposed algorithm.