With the fast development of storage technologies, large-scale and high dimensional datasets are stored in a distributed way. It usually applies distributed clustering algorithms to cluster distributed datasets. This paper presents a distributed clustering algorithm based on Clique and high dimensionality reduction to do the distributed clustering. Moreover, the efficiency, accuracy and extendibility of clustering analysis are improved by self-adapting algorithms and the assistant of data and mission parallelism in master or child node. Through experiments, we show that DPA-CLIQU efficiently finds accurate clusters in large high dimensional datasets from a distributed system.
With increasing needs of a live video streaming service to mobile devices, peer-to-peer video delivery is required to provide seamless quality of video even if a user moves across heterogeneous networks. In this paper, we address technical challenges and approaches to providing users a seamless video quality in P2P-based live video streaming services over mobile heterogeneous networks by investigating on the intrinsic characteristics of the underlying technologies.
This research aims to develop support methods for grasping discussion transition processes from e-mails. This research has proposed algorithms for extracting the streams of discussions and their progress states in utterance level. This paper mainly describes a support system for visually extracting the discussion processes. This paper also describes the experiments focusing on method replaceability in each extraction process together with the outlines of the algorithms and the prototype. Finally, the features of the methods are discussed based on the results.
This research describes the design of a virtual architectural walkthrough application, developed and run on two different platforms; Windows and Linux operating system. Both platforms are then compared in order to measure and evaluate the performance of each platform used to build the virtual architectural walkthrough. Therefore, this research is an attempt to quantify the relative merits of the Windows and Linux operating systems as an underlying platform for a virtual architectural walkthrough and to experimentally determine, which either of the two demonstrates superiority for this task. This is to make sure that the base in which the particular application is run at has sufficient help at its maximum performance. This research is also important to show which operating system has better capabilities, speed and ample of memory to sustain a virtual reality application. In pursuit of this goal, several experiments are detailed and recorded, in which key criteria are compared between the two operating systems. The performance of each platform is measured based on four main key criteria's which is frame rate, image quality, CPU usage and memory usage.
We present a demonstrating Web application for biomedical imaging and visualization, addressing its capability of implementing comprehensive image processing algorithms at the server side where proprietary datasets reside and direct access to them by client-side applications is restricted. The main technical challenge is to make the functionality of the dedicated imaging and visualization programs (usually compiled from a native programming language such as C++) publicly available via a Web based interface. Accessing information of the original image data and algorithmic details should be hidden, thus scripts executed by the webpage browser should be avoided. In this regard, we propose a solution based on the ASP.NET framework. Managed codes in C# are developed for Web controls, server-side GUI and relational database management; whereas unmanaged server-dependent imaging programs are invoked from within the managed programs. The advantages of such a Web application include (a) high security of the proprietary data which is always a primary consideration in deployment of a networked system, (b) thin-client strategy in which heavy computing tasks are completed on the dedicated server with high-performance hardware, and (c) less demand for network bandwidth as only the rendering results instead of multiple high-dimensional datasets need to be transferred. The proposed approach is believed to be beneficial for building the Web-based Laboratory Information System (LIS) and Radiology Information System (RIS).
We report an approach to estimate the functional connection from the possible anatomical connections and physiological response of each neuron. Here we focus on the brain of Bombyx mori, a male silkmoth, which exhibits a programmed behavior for the pheromone orientation on a pheromone detection. The neuroethological experiments revealed that the motor command is generated in the premotor center, composed of the lateral accessory lobe (LAL) and the ventral protocerebrum (VPC). The morphology and the physiological response to the pheromone stimulus are obtained for each main type of LAL-VPC neuron through the intensive electro-physiological and immunohistochemical experiments. However, how the motor command is generated in LAL-VPC, or the information pathway in the network, is still unknown. That is, each piece of puzzle is fairly well investigated, though they are not yet fit in to reveal the whole picture. The goal of our present study is to investigate the functional connection in LAL-VPC to elucidate the mechanism of the motor command generation for the programmed behavior. The present approach is to consider an artificial three-layered neural network to solve an optimization problem based on the above findings of LAL-VPC neuron as the constraints. The optimal solution is the connection which reproduces the observed response best, and we search it through the learning of the artificial neural network. Then, the obtained connection is evaluated by a simulation of LAL-VPC network to reproduce the network dynamics.
The standardization of IEEE 802.15.4g SUN has been proceeded to provide a global standard that facilitates very large scale process control applications such as the utility smart-grid network. SUN devices are designed to operate in very large-scale, low power wireless applications in order to support the data communications network for smart grid. In this paper, We estimate the performance of the narrowband PHYs for SUN devices considering the RFID channel spacing between 917 MHz and 923.5 MHz in Korea Unlicensed band.
Digital costume is clothing for performer which consists of lighting devices and clothing materials. It can provide stronger visual effects than traditional performer costume. Because the digital costume is hard to modify after built, design of the digital costume is one of the most important procedures. This paper proposes Digital costume visualizer, which visualizes digital costume in 3D virtual space. The visualizer creates virtual digital costume based on the costume model and luminous pattern which created by a user. To create a visual appearance which similar to the real digital costume we applied Bloom effect to the luminous patterns. By applying several control values to the Bloom effect, the visualizer can provide various visual effects. As a result, the visualizer provides enough functionalities and effects for design process of the digital costume.
To inquire into the change of the concentration of inflammatory factors (TNF-a, IL-1, IL-6) in blood plasma of rats with myocardial ischemia, which was the syndrome of Qi deficiency and Blood Stasis syndrome. The accompanying cardiac function also be investigated, relationship between them are analyzed. Rats were divided into either LAD surgery group or sham operation group randomly. Syndromes associated physical and chemical indexes were measured at 4, 7, 28, 45 and 60 days after surgery. Compared with sham operation group, The concentration of IL-1, IL-6 and TNF-a in model group all obviously changed, EF, FS changed obviously, which decreased by 63.44% and 76.67% respectively, and the express of TNF-α and LVEDs, LVEDd presented negative correlation, with the deteriorate of cardiac function, it increased. We can conclude that in earlier stage of the Syndrome of Qi deficiency and blood stasis, inflammatory cytokines especially TNF-α play a vital role that contribute the pathology mechanism, TNF-α which has a negative correlation with cardiac function, can be used as indicators of inflammatory activation and can reflect the state of cardiac function. Current data indicate that plasma levels of inflammatory markers especially TNF-α may have clinical utility in helping identify and cure individuals at high risk for future blood stasis syndrome.
We improved RANSAC filtering with similarity between matching pairs of local features. As knowns, local features, such as SIFT, use descriptor for matching between different feature set. The method of distance ratio of nearest neighbors are employed to judge whether a point pair is an acceptable matching pair or not. Our proposed similarity is determined by distance ratio of nearest neighbors among matching pairs. An ordered queue of similarities corresponding to matching pairs is built up. The matching pairs in front part of the similarity queue are utilized in a linear-order combination to compute the transformation matrix. Experiments show that the proposed method can find more inliers than original RANSAC and can generate an accurate transformation matrix in a few iterations. The advancement of the proposed method is due to the priori knowledge of similarity of matching pairs.
In this work, a dynamic model of S. vulgaris shrubs under drought stress was discussed based on improved dual-scale automaton model with the data from experiments. Relationship of groundwater level and growth of S. vulgaris has been confirmed by computing the growth rate of tree height and radius growth on drought. The growth probability has been given with segment function in different condition. Effect of growth has been analyzed in essential growth cycle by graph. Comparing with real S. vulgaris, the result indicates that the model can characterize the growth of S. vulgaris more truly. The visualization indicates that the model is highly consistent with the growth of real S. vulgaris under drought stress.
This paper proposed a novel Hilbert-Huang transform (HHT) based endpoint detection method. Firstly, IMF1 and IMF2 are discarded after EMD, and then some IMFs components that can be used in endpoint detection are selected to reconstruct the speech signal by comparing means and variances. The reconstructed signal is used for detection via two-threshold method. The experiences show that the proposed method remains high detection rate of 84.2% compared with two-threshold method. When there are noises existing in speech signal, the proposed method still can implement detecting with detection rate being 85.9% while traditional two-threshold method is invalid. All these demonstrate that the new approach has robustness in some degree.
In this paper, we propose a simple, fast and accurate content extraction method: CELB. Compared with traditional methods, this approach does not parse the DOM trees and uses only information from lines of original HTML documents. We propose a concept called line-block, to extract contents more effectively and a new feature distance-text number (DTN) for distinctions between contents and non-contents. First, we preprocess original HTML documents, and then combine lines into line-blocks. Next, we calculate values of content features for each line-block, and use thresholds to determine whether a lineblock is part of the main content or not. Experiments show satisfied results, especially for the running time.
Support vector machine is a widely used tool in the field of image processing and pattern recognition. The parameters selection plays a significant role in support vector machine(SVM). This paper proposed an improved parameter optimization method based on traditional PSO optimizing algorithm by changing the fitness function in the traditional process. And this method has achieved better results which reflected in the ROC curves in medical images classification.
By taking 223 data from 14 famous telecom operators all of the world as sample, this research investigates the mechanism between factors (which reveals the characters of telecom operation industry) and Economic Value Added. The methodologies mainly used include Factor analysis, Correlation Analysis, and Structural equation modeling (SEM). Final result indicates that the factors affecting EVA of telecom operator mainly includes Profitability capacity, Management capacity, and Risk Resist capacity, which positively affect EVA. Scale and Growth speed are not significant in improving telecom operator's EVA. Result shows that managers of telecom operators should pay more attention to the quality and health of companies.
This paper presents a new approach to the automatic detection of worms using behavioral signature with sequential hypothesis testing. Characteristic patterns of worm behaviors in network traffic include 1) sending similar data from one machine to the next, 2) tree-like propagation and reconnaissance, and 3) changing a server into a client [1]. 4) infected hosts should meet the basic requirements [2]. This sequential hypothesis testing theory first developed by Wald [3] overcomes the hurdle that it's difficult to find a threshold of determining whether a host is infected when we build a tree of worm propagation. Besides, to slow down the propagation of worms we alter an existing method which is connection rate limiting to packet-sending rate limiting. Our results that show this approach successfully restricts the number of packets sent by worms, has a low false alarm rate and a low missing alarm rate.
Mutual information is a widely-used metric in medical image registration, which achieves many good results; however, it may lead to misregistration because of the absence of spatial information. In order to solve this problem, this paper proposes an improved hybrid metric combining mutual information and spatial information. The spatial information is obtained through calculating Hausdorff fraction. In experiments, mutual information is used as a comparative method. Experiments show that the proposed method performs well in multiphase contrast-enhanced abdomen CT image non-rigid registration problems.
Combined Input-Crosspoint-Queued (CICQ) crossbar switches are advantageous in performance over input-queued (IQ) switches. As a result, many scheduling schemes have been developed for CICQ switches to further improve their performance. This paper presents a new scheduling scheme called SBF-GWF (the Shortest buffer First and the Greatest Weigh buffer First). It takes the advantage of the SBF algorithm used in MCBF input arbiters with an additional feature that uses VOQ lengths as weights, which will in turn be used for privileged output with the fluid model, we have proved that SBF-GWF algorithm can achieve 100% throughput for any admissible traffic without the need of speed-up for crossbars. Our simulation results have also shown that SBF-GWF algorithm exhibits good delay performance under any admissible traffic compared with existing scheduling schemes.
To obtain the benefit of aggressive CMOS scaling, chip multiprocessors (CMP) has the tendency of integrating more processors on one chip. However, the increase of cores and voltage factors are increasing susceptibility of CMP to transient faults, hard errors, and process variations. This challenge makes thread-level redundancy (TLR) more attractive in fault tolerance. The proposed CMP architecture executes two copies of a thread on separate cores, so one hard error cannot influence both threads simultaneously. Redundant threads maintain identical hardware resources and independent memory operations. The committed store can update L1 cache before verification. Because updating the unverified values to L2 cache is not allowed, redundant threads do not suffer the non-determinism in parallel applications. Added cache architecture is proposed to buffer thousands of unverified instructions; consequently long latencies in verification can be tolerated. The experimental results show that, for several parallel applications from the SPLASH-2, the performance loss incurred is only 3% due to redundancy.
