Abstract: In the design and application of embedded Linux operating system,operating system runs on different hardware platforms after transplantation,and it needs a task scheduling algorithm for effective implementation of process management and memory management to improve the operational efficiency of the system.Linux system dual threshold scheduling algorithm was proposed based on characteristic scale equilibrium.The kernel structure of embedded Linux was analyzed.The system task scheduling model was constructed.According to the various classifications of information such as task arrival rate,execution time,etc,the scale features are extracted.In the global task scheduling center,all the task data are integrated and input to the total system scheduler,and the scale optimization objective function is obtained.The feature scale balanced processing is taken.The characteristic time shaft is divided into the adjacent but not overlap task matching smoothing window,and the double threshold trade-off decision is used for task scheduling in Linux system.The simulation results show that the new algorithm has higher efficiency in Linux embedded task scheduling,utilization rate of CPU is better,and the overall performance is better than the traditional algorithm.

Abstract: R&D in the nuclear reactor physics demands state-of-the-art numerical tools that are able to characterize investigated nuclear systems with high accuracy. In this paper, we present the Monte Carlo Continuous Energy Burnup Code (MCB) developed at AGH University’s Department of Nuclear Energy. The code is a versatile numerical tool dedicated to simulations of radiation transport and radiation-induced changes in matter in advanced nuclear systems like Fourth Generation nuclear reactors. We present the general characteristics of the code and its application for modeling of Very-High-Temperature Reactors and Lead-Cooled Fast Rectors. Currently, the code is being implemented on the supercomputers of the Academic Computer Center (CYFRONET) of AGH University and will soon be available to the international scientific community.

Abstract: Semantic similarity measures play important roles in many Web-related tasks such as Web browsing and query suggestion.Because taxonomy-based methods cannot deal with continually emerging words,recently Web-based methods have been proposed to solve this problem.Because of the noise and redundancy hidden in the Web data,robustness and accuracy are still challenges.We proposed a method integrating page counts and snippets returned by Web search engines.Then,the semantic snippets and the number of search results were used to remove noise and redundancy in the Web snippets.After that,a method integrating page counts,semantics snippets and the number of already displayed search results was proposed.The proposed method does not need any human annotated knowledge,and can be applied Web-related tasks easily.A correlation coefficient of 0.851 against Rubenstein-Goodenough benchmark dataset shows that the proposed method outperforms the existing Web-based methods by a wide margin.Moreover,the proposed semantic similarity measure significantly improves the quality of query suggestion against some page counts based methods.

Abstract: In complex networks,how to rank the nodes according to their importance plays key roles in various kinds of fields.Against the problem that most of the existing single index are unilateralist and limited,and most of the current synthesis evaluation method is also inaccurate,we proposed a new synthesis evaluation method combining principal component analysis with TOPSIS.We used our method to analyse the ARPA and the USA airport network.The result suggests that our method should be effective and corrective,and also lays the foundation for further more evaluation of node importance.

Abstract: In this paper, we analyze existing quantum computer simulation techniqu- es and their realizations to minimize the impact of the exponential complexity of simulated quantum computations. As a result of this investigation, we pro- pose a quantum computer simulator with an integrated development environ- ment - QuIDE - supporting the development of algorithms for future quantum computers. The simulator simplifies building and testing quantum circuits and understanding quantum algorithms in an ecient way. The development envi- ronment provides flexibility of source code edition and ease of the graphical building of circuit diagrams. We also describe and analyze the complexity of algorithms used for simulation as well as present performance results of the simulator as well as results of its deployment during university classes.

Abstract: Protection of infrastructures for e-science, including grid environments and NREN facilities, requires the use of novel techniques for anomaly detection and network monitoring. The aim is to raise situational awareness and provide early warning capabilities. The main operational problem that most network operators face is integrating and processing data from multiple sensors and systems placed at critical points of the infrastructure. From a scientific point of view, there is a need for the efficient analysis of large data volumes and automatic reasoning while minimizing detection errors. In this article, we describe two approaches to Complex Event Processing used for network monitoring and anomaly detection and introduce the ongoing SECOR project (Sensor Data Correlation Engine for Attack Detection and Support of Decision Process), supported by examples and test results. The aim is to develop methodology that allows for the construction of next-generation IDS systems with artificial intelligence, capable of performing signature-less intrusion detection.

Abstract: Research described in this paper tries to combine the approach of Deep Neural Networks (DNN) with the novel audio features extracted using the Scatter- ing Wavelet Transform (SWT) for classifying musical genres. The SWT uses a sequence of Wavelet Transforms to compute the modulation spectrum coef- ficients of multiple orders, which has already shown to be promising for this task. The DNN in this work uses pre-trained layers using Sparse Autoencoders (SAE). Data obtained from the Creative Commons website jamendo.com is used to boost the well-known GTZAN database, which is a standard bench- mark for this task. The final classifier is tested using a 10-fold cross validation to achieve results similar to other state-of-the-art approaches.

Abstract: In this paper, we present the findings of a qualitative analysis of 15,750 com- ments left by 2,041 participants in a Reconcile web credibility evaluation study. While assessing the credibility of the presented pages, respondents of the Re- concile studies were also asked to justify their ratings in writing. This work attempts to give an insight into the factors that aected the credibility asses- sment. To the best of our knowledge, the presented study is the most-recent large-scale study of its kind carried out since 2003, when the Fogg et al. -How do users evaluate the credibility of Web sites? A study with over 2,500 parti- cipants' paper was published. The performed analysis shows that the findings made a decade ago are still mostly valid today despite the passage of time and the advancement of Internet technologies. However we report a weaker impact of webpage appearance. A much bigger dataset (as compared to Fogg's stu- dies) allowed respondents to reveal additional features, which influenced the credibility evaluations.

Abstract: The glowworm swarm optimization algorithm(GSO)or firefly algorithm(FA)is one of the intelligence algorithms,which is inspired by the biological behavior of the glowworm attracting mates or preyingIt has good performance in the discrete combinational optimization problems and continuous optimization problemsHowever,it still has some drawbacks such as it's easily trapped into local optimal solutionsStarting with the improvement and fusion of the algorithm as well as the discrete mechanism,this paper presented a series of schemes on improving the GSO or FAFinally,some meaningful remarks on the future research were presented

Abstract: Web attack detection is a kind of dynamic Web security protection technology,but the intruder can use different coding schemes,mixed case,alternative statements and other skills,bypassing defense mechanism.For the particularity of web security and the shortage of the existing detection technology,we took SQL injection and cross site scripting attacks as an example.Firstly,the thesis studies the feature selection and extraction of SQL injection and cross site scripting attacks,and uses the artificial selection and mathematical statistical methods to covert the original payload into fixed dimension feature vector.Secondly,it marks the sample data after feature selection and extraction,and performs support vector machine training and classification.Finally,using the Weka,it verifies the feasibility and effectiveness of the approach.The experimental results show that features after selection and extraction can reflect the nature of the original data and this method has higher detection rate.

Abstract: According to the optimization capability of evolutionary algorithms,we proposed an object-oriented remote sensing image classification method based on GEPSO(GEP Optimized by PSO,GEPSO)modelFirstly,image segmentation was done,feature set was selected for the remote sensing image,and then uses GEPSO algorithm was used to construct a class center for each type of image objectsThe process of constructing class centers firstly makes use of GEP to search a suboptimal solution,and then uses PSO to search the optimal solution with the suboptimalExperimental results show that the object-oriented remote sensing image classification method based on GEPSO model has higher classification accuracy

Abstract: As the online shopping is becoming more and more popular,buying medicine online has brought great convenience for many patientsBut when ordinary people buy drugs online,they always purchase medicine blindlyThere is a big problem that they do not have access to the medicine guidanceIn order to solve this problem,firstly,we clustered the drug into several groups according to the functional description information of the drug,and proposed the personalized medicine recommendation based on user collaborative filteringThen considering the shortcomings of the collaborative filtering algorithm,we used the tensor decomposition methods to model the relationship of the user,symptom and medicine,and recommended the top-N related medicines to the users according to their symptomsWe crawled the real data from the internet and compared the results with collaborative filtering methodThe results show good performance

Abstract: Deep neural networks are often trained in two phases: first hidden layers are pretrained in an unsupervised manner and then network is fine-tuned with error backpropagation. Pretraining is often carried out using Deep Belief Networks (DBNs), with initial weights set to small random values. However, recent results established that well-designed initialization schemes, e.g. Sparse Initialization (SI), can greatly improve performance of networks that do not use pretraining. An interesting question arising from these results is whether such initialization techniques wouldn't also improve pretrained networks? To shed light on this question, in this work we evaluate SI in DBNs that are used to pretrain discriminative networks. The motivation behind this research is our observation that SI has an impact on the features learned by a DBN during pretraining. Our results demonstrate that this improves network performance: when pretraining starts from sparsely initialized weight matrices networks achieve lower classification error after fine-tuning.

Abstract: Vibe algorithm is an effective pixel level background modeling algorithm.During the moving object detection process,the Vibe algorithm can't eliminate ghost quickly and change the update speed according to the change speed of foreground.To solve these problems,this paper proposed a threshold based adaptive Vibe target detection algorithm.When a pixel is judged as foreground by Vibe model,Otsu algorithm will be used to calculate the threshold of image segmentation.The pixel will be judged again to eliminate ghost pixel according to the threshold,and the background model of the pixel is reinitialized.According to calculate the change of the centroid of the moving object,the improved algorithm changes the update rate of the background adaptively.The results show that the proposed algorithm can effectively absorb the ghost in less frames and detect the foreground object more accurately than the original Vibe algorithm.

Abstract: The increasing availability and popularity of computer systems has resulted in a demand for new language- and platform-independent ways of data exchange. This demand has, in turn, led to significant growth in the importance of systems based on Web services. Alongside the growing number of systems accessible via Web services came the need for specialized data repositories that could oer eective means of searching the available services. The development of mobile systems and wireless data transmission technologies has allowed us to use dis- tributed devices and computer systems on a greater scale. The accelerating growth of distributed systems might be a good reason to consider the develop- ment of distributed Web service repositories with built-in mechanisms for data migration and synchronization.

Abstract: This paper presents new trends in interface design of medical ultrasound devices and explains the basics of implementing usability engineering in accordance with international standards. Methods for determining the initial requirements, design guidelines, processes of prototyping, verification, and validation of software user interfaces for medical devices are discussed. The article also presents a preliminary plan of a methodology for prototyping touch-based and standard-control interfaces for mobile ultrasonic devices.

Abstract: Rough set is one of the basic methods in dealing with the imprecise or indefinite problems.For its advantages that the priori knowledge about analyzing dataset isn't necessary and the parameters analysis needn't to be set artificially,rough set is widely used in pattern recognition and data mining fields.For rough set theory,a core problem is how to classify the sample which has never been met in the process of training.This problem was discussed in detail in this paper.According to the importance of the condition attributes,a weighted KNN algorithm was proposed to classify the samples which can't precisely match to decision rules,and the contrast test with the weighted minimum distance(WMD)method was made to show the efficiency of our algorithm.At the same time,the existing algorithms about the attribute value reduction in rough set were analyzed and another point of view was put forward.The experiments on several UCI data sets and comparison with various existing algorithms proposed recently show that our algorithm is superior to these algorithms in overall effect.

Abstract: Dispute and controversy are parts of our culture and cannot be omitted on the Internet (where it becomes more anonymous). There have been many studies on controversy, especially on social networks such as Wikipedia. This free on-line encyclopedia has become a very popular data source among many researchers studying behavior or natural language processing. This paper presents using the category structure of Wikipedia to determine the controversy of a single article. This is the first part of the proposed system for classification of topic controversy score for any given text.

Abstract: Hate speech and fringe ideologies are social phenomena that thrive on-line. Members of the political and religious fringe are able to propagate their ideas via the Internet with less eort than in traditional media. In this article, we attempt to use linguistic cues such as the occurrence of certain parts of speech in order to distinguish the language of fringe groups from strictly informative sources. The aim of this research is to provide a preliminary model for iden- tifying deceptive materials online. Examples of these would include aggressive marketing and hate speech. For the sake of this paper, we aim to focus on the political aspect. Our research has shown that information about sentence length and the occurrence of adjectives and adverbs can provide information for the identification of dierences between the language of fringe political groups and mainstream media.

Abstract: Parkinson's Disease (PD) is primary related to substantia nigra degeneration and, thus, dopamine insuciency. of stimulation related to the interruption of pathological oscillation in the basal ganglia found in PD. Our model represents possible STN neural population with inhibitory and excitatory connections that have patho- logically synchronized oscillations. High-frequency electrical stimulation has interrupted synchronization. something that is also observed in PD patients.

Abstract: The contradiction of comprehensive information and dimension curse is the preliminary problem of network situation awareness in the times of big data.Feature extraction is a mainstream method to dimensionality reduction,but performs not well when solving high-dimension and nonlinear data.Deep learning is a multi-layer and nonlinear algorithm which can realize the approximation of complicated function,however,it is sensitive to parameters related to hidden layer.Based on above analysis,a feature extraction algorithm based on evolutionary deep learning was proposed.The algorithm combines evolutionary algorithm(EA)and deep learning,takes advantage of the characteristics of GA and ES,and optimizes the learning structure and relevant parameters.Theoretical analysis and simulation results both prove the effectiveness of this algorithm.

Abstract: This paper describes the application of hypergraph grammars to drive a linear computational cost solver for grids with point singularities. Such graph gram- mar productions are the first mathematical formalisms used to describe solver algorithms, and each indicates the smallest atomic task that can be executed in parallel, which is very useful in the case of parallel execution. In particular, the partial order of execution of graph grammar productions can be found, and the sets of independent graph grammar productions can be localized. They can be scheduled set by set into a shared memory parallel machine. The graph- grammar-based solver has been implemented with NVIDIA CUDA for GPU. Graph grammar productions are accompanied by numerical results for a 2D case. We show that our graph-grammar-based solver with a GPU accelerator is, by order of magnitude, faster than the state-of-the-art MUMPS solver.

Abstract: Cloud-P2 Pcombines the resources of all nodes of cloud computing and peer-to-peer computing to achieve the largest collaboration and resource sharing.The data destruction mechanism is one of the important measures to protect users' data security and controllability,which is difficult for Cloud-P2 Psystems.In order to meet the requirement of data destruction in Cloud-P2 Pstorage systems,a composite data destruction mechanism based on mobile agent was put forward,which can make the expired,waste data destructed effectively,as well as defend those malicious attacks on data.In order to effectively destruct data on one node with low cost,a novel data destruction method was proposed,which realizes the data destruction by data folding.

Abstract: After developing a Resonance Chiral Lagrangian (RχL) model to describe hadronic τ lepton decays [18], the model was confronted with experimental data This was accomplished using a fitting framework which was developed to take into account the complexity of the model and to ensure the numerical stability for the algorithms used in the fitting Since the model used in the fit contained 15 parameters and there were only three 1-dimensional distributions available, we could expect multiple local minima or even whole regions of equal potential to appear Our methods had to thoroughly explore the whole parameter space and ensure, as well as possible, that the result is a global minimum This paper is focused on the technical aspects of the fitting strategy used The first approach was based on re-weighting algorithm published in [17] and produced results in around two weeks Later approach, with improved theoretical model and simple parallelization algorithm based on Inter-Process Communication (IPC) methods of UNIX system, reduced computation time down to 2-3 days Additional approximations were introduced to the model decreasing time to obtain the preliminary results down to 8 hours This allowed to better validate the results leading to a more robust analysis published in [12]

Abstract: In the process of research on complex networks,we divided its network structure into three types according to the node connection tendency,including disassortative network,assortative network and the neutral network.We adopt variable gradient analysis to judge and analyze it's stability respectively.Theoretical analysis shows that disassortative network is stable within a wide range,the assortative network's state is unstable,and the stability of the neutral network is uncertain.We can determine whether the network is in a steady state based on the tendency of node connection.Meanwhile,in the research on complex network robustness,related simulation results show the stability and robustness have positive correlation,and the robustness of the disassortative network is best and the secondary is neutral network,followed by the fragile robustness of assortative network.

Abstract: Current static-analysis practice on Android application package(APK)mainly uses the features such as permissions,data flows,API calls,extracted from the manifest file and the code.Such features lack consideration on the APK code organizations and object hierarchy,and thus they may be ineffective in describing and predicting an APK's application behaviors and maliciousness.This research work tried to design and implement a comprehensive API-usage characterization method for Android APK on different resolutions and hierarchies,namely packages,classes,and functions.A tree structure is used to contain such hierarchical API-usage information,and a comparison algorithm is designed for cross-tree similarity,which provides extra insights in classifying and differentiating Android malware of different types and code families.The variations in API-usage on different code layers imply code functionalities and application behaviors,and thus they can be used to improve current static-analysis based malware detection and signature generation.Realistic malware packet samples of various types and families were used to validate the proposed characterization method,and results were discussed for its strength and future improvement.

Abstract: The equation for nonlinear diffusion can be rearranged to a form that immediately leads to its stochastic analog. The latter contains a drift term that is absent when the diffusion coefficient is constant. The dependence of this coefficient on concentration (or temperature) is handled by generating many paths in parallel and approximating the derivative of concentration with respect to distance by the central difference. This method works for one-dimensional diffusion problems with finite or infinite boundaries and for diffusion in cylindrical or spherical shells. By mimicking the movements of molecules, the stochastic approach provides a deeper insight into the physical process. The parallel version of our algorithm is very efficient. The 99% confidence limits for the stochastic solution enclose the analytical solution so tightly that they cannot be shown graphically. This indicates that there is no systematic difference in the results for the two methods. Finally, we present a direct derivation of the stochastic method for cylindrical and spherical shells.

Abstract: As a distributed cloud platform,Hadoop is one of the most widely used cloud storage technology for applications with large datasets to provide reliable and efficient storage service,but it suffers a performance penalty with increased number of small files.In order to improve the efficiency of storing and accessing the small files on Hadoop,we proposed a scheme,based on the relationship of small files.In the scheme,a set of correlated files is combined into a large file to reduce the file count,indexing mechanism is used to access small file and metadata cache,and associated small file prefetching mechanism is used to improve the efficiency of file read.The experimental results indicate that the above methods can improve the storage and access efficiency of small file on Hadoop.

Abstract: This paper describes a method for geolocalized information retrieval from natural language text and its interpretation by assigning them geographic coordinates. A proof-of-concept implementation is discussed, along with geolocalized dictionary stored in PostGIS / PostgreSQL spatial relational database. Discussed research focuses on strongly inflectional Polish language, hence additional complexity had to be taken into account. Presented method has been evaluated with the use of diverse metrics.