Jianpei Zhang

An Improved Hierarchical Multi-class Support Vector Machine with Binary Tree Architecture

A hierarchical binary tree multi-class support vector machine (BTMSVM) based on class similarity in feature space is improved to overcome the drawbacks such as unclassifiable region which the existent methods have. The class similarity which considers class distance and distribution sphere in feature space is used to determine the classification order of hierarchical multi-class SVM. The learning samples and corresponding SVM sub-classifier are selectively re-constructed to make sure as bigger as classification margin, as much as generalization ability. The results of simulated experiments show that the proposed method is faster in training and classifying, better in classification correctness and generalization.

Software reliability growth model based on fuzzy wavelet neural network

Back-propagation neural network (BPNN) for software reliability prediction has the problem of network architecture difficult to determine. We want to solve the problem and further increase the prediction accuracy of the model. We use phase space reconstruction technology and the fuzzy wavelet neural network (FWNN), and present the software reliability growth model based on fuzzy wavelet neural network (FWNN-SRGM). The experimental results show that the presented method can easily determine the network architecture according to failure data. The prediction accuracy of FWNN-SRGM is better than that of the widely used BPNN-SRGM.

Dynamic density-based clustering algorithm over uncertain data streams

In recent years, the uncertain data stream which is related in many real applications attracts more and more attention of researchers. As one aspect of uncertain character, existence-uncertainty can affect the clustering process and results significantly. The lately reported clustering algorithms are all based on K-Means algorithm with the inhere shortage. DCUStream algorithm which is density-based clustering algorithm over uncertain data stream is proposed in this paper. It can find arbitrary shaped clusters with less time cost in high dimension data stream. In the meantime, a dynamic density threshold is designed to accommodate the changing density of grids with time in data stream. The experiment results show that DCUStream algorithm can acquire more accurate clustering result and execute the clustering process more efficiently on progressing uncertain data stream.

Research on cultural algorithm for solving routing problem of mobile agent

Abstract The key idea behind cultural algorithm is to explicitly acquire problem-solving knowledge from the evolving population and in return apply that knowledge to guide the search. In this article, cultural algorithm-simulated annealing is proposed to solve the routing problem of mobile agent. The optimal individual is accepted to improve the belief spaces evolution of cultural algorithms by simulated annealing. The step size in search is used as situational knowledge to guide the search of optimal solution in the population space. Because of this feature, the search time is reduced. Experimental results show that the algorithm proposed in this article can ensure the quality of optimal solutions, and also has better convergence speed. The operation efficiency of the system is considerably improved.

A K-anonymity clustering algorithm based on the information entropy

Data anonymization techniques are the main way to achieve privacy protection, and as a classical anonymity model, K-anonymity is the most effective and frequently-used. But the majority of K-anonymity algorithms can hardly balance the data quality and efficiency, and ignore the privacy of the data to improve the data quality. To solve the problems above, by introducing the concept of “diameter” and a new clustering criterion based on the parameter of the maximum threshold of equivalence classes, we proposed a K-anonymity clustering algorithm based on the information entropy. The results of experiments showed that both the algorithm efficiency and data security are improved, and meanwhile the total information loss is acceptable, so the proposed algorithm has some practicability in application.

Research on cooperation communication based-on cross-layer for underwater wireless sensor networks

With the increase of requirements for the ocean information in the world, people begin to put focus on the research on underwater wireless sensor networks. The research on underwater wireless sensor networks is studied by the layer methods in the wireless sensor networks. Although this simplifies the research, the respective layer optimization is not able to guarantee the overall networks optimization. Cross layer can make up for the limitation brought by layer. Its object is to realize the balance between the performance and the design interaction in protocol layer which is not adjacent in logicalness. The cross layer among the physical, data link and networks balances the performance and decreases the transition about the control signal. It realized the maximization of saving energy and the extension of lifetime. This is very important to the underwater wireless sensor networks. The paper combines underwater communications among underwater devices, takes use of the cross layer design, considers the physical layer, modulation and forward error correct interaction, media access control and routing interaction. Based on the research on wireless sensor networks, combined with the characteristics of physical layer, the functionalities and characteristics are considered in a comprehensive way, the cross layer design is studied thoroughly, and the cross layer link transmission algorithm and cross layer link optimization are proposed which makes the scare resource effective utilization. Simulation experiments show that cooperation communication based-on cross layer is fit for underwater wireless sensor network. The algorithm increase the efficiency and throughout. They also have good stability.

A Survivability Information System Based on Service Self-Organization

A survivability information system based on service self-organization is proposed under the precondition of modules redundancy backup. The service self-organization way divides the system into services communications modules, services distribution modules, data storage modules according to service processing flow, the backup modules send its survivability information to the working modules duly, and get the latest service states list. When a module has the highest survivability, the system will do the service self-organization through different connection handoff techniques according to the functions of the modules and their location in system architecture. The validity of the survivability calculation method is proved by the simulation experiment. Examination proved that the method proposed in this paper can enhance the service survivability both of the steady state and instantaneous state by setting up a Web server in the network environment.

Research on data streams publishing of privacy preserving

Data streams contain a lot of client information that need to be carefully managed to protect privacy of clients. Most of existing privacy preserving methods, such as k-anonymity, was designed for static data sets. However these methods can not be applied on data streams directly. Moreover, in dada streams applications, there is a need to offer strong guarantees on maximum allowed delay between incoming data and its anonymous output. This paper presents a novel weak clustering based data streams k-anonymity method for data publishing. Three advantages make its practical: first, little processing time for each tuple of data steam. Second, demand less memory. Last, both privacy preserving and utility of anonymous data are considered carefully. Theoretical analysis and experimental results show that the method is efficient and effective.

A Content-Based Self-Feedback E-government Network Security Model

Based on the theories of the intrusion trapping and natural language understanding, oriented e-government affairs security issues, this paper proposed a content-based self-feedback model at the point of attackers. By this model, the concrete information under attacking can be focused and the attack methods would be ignored in a standard honey trap. With the supporting of honey nets, The target sensitivity is taken as the appraisement factor to demonstrate whether the information need be protected. Through adjusting the primitive feedback coefficients of the model, we can know the most important information as the attackers focusing on. At the same time, this paper introduced the concept of domain coefficients of the model. Through the experiments in the actual networks, it is the first successful model being of the prediction and feedback for e-government affairs.

Research on Semantic Web-Oriented Ontology Model

The World Wide Web is an open and distributed repository with lots of information, in which the ontologies can help users to get their required information. This paper improves representation method of the existing ontology by introducing matter-element and its extensibility in extenics, and put forward a new ontology model-extenics knowledge ontology model. In this paper, some research has been done in order to make semantic Web stronger in network information search and mining.