Abstract:It is proved that given any nonrecursive r.e. degree a, there exist r.e. degrees c

Abstract:Sign language is the language used by the deaf, which is a comparatively s teadier expressive system composed of signs corresponding to postures and motion s assisted by facial expression. And it is a language communicated by motion/vis ion. The objective of sign language recognition research is to “see” the langu age of the deaf. The integration of sign language recognition and sign languag e synthesis jointly comprises a “human-computer sign language interpreter”, w hich facilitates the interaction between deaf people and their surrounding s. The issue of sign language recognition is to recognize dynamic gesture signal , that is, to recognize sign language signal. Considering the speed and performa nce of the recognition system, Cyberglove is selected as gesture input device in sign language recognition system, DGMM (dynamic Gaussian mixture model) is used as recognition technique, and hierarchical recognizer is used in recognizing mo dule, which can recognize 274 sign language words coming from the dictionary of Chinese sign language with an accuracy of 97.4%, based on Chinese sign language' s own characteristic. Compared with the recognition system based on single-DGMM recognizer, the recognition rate of hierarchical DGMM recognizer is nearly equa l to that of single-DGMM recognizer, but its recognition speed is apparently mu ch faster than that of single-DGMM recognizer.

Abstract:In this paper, a generalize d fuzzy CMAC (cerebellar model articulation controller) is presented, the suffici ent condition of the learning of the generalized fuzzy CMAC is derived, and fina lly the learning convergence of the generalized fuzzy CMAC to the least square error is proved. The results provide a mathematical foundation for the generaliz ed fuzzy CMAC's wide applications.

Abstract:In this paper, a Field Theory based adaptive resonance neural network algorithm FTART, which combines the advantages of Adaptive Resonance Theory and Field Theory, is proposed. FTART employs a unique approach to solve the conflicts between instances and extend classification regions dynamically. So that it does not need user to manually configure hidden units, and achieves fast training speed and high predictive accuracy. Moreover, a method named Statistic based Producing and Testing, which has the ability of extracting comprehensive and accurate symbolic rules from trained FTART,is proposed.Experimental results show that the symbolic rules extracted via this method can commendably describe the function of FTART.

Abstract:Intrusion detection is a highlighted topic of network security research in recent years. In this paper, first the necessity o f intrusion detection is presented, and its concepts and models are described. T hen, many intrusion detection techniques and architectures are summarized. Final ly, the existing problems and the future direction in this field are discussed.

Abstract:A new random CSP (constraint satisfaction pro blem) model is proposed in this paper. By analyzing the expected number of nodes in a search tree, the average running time used by the backtracking algorithm o n random constraint satisfaction problems is studied. The results show that the model can generate hard CSP instances, and the expected number of nodes required for finding all solutions or proving that no solution exists becomes exponentia lly large as the number of variables grows. Therefore, the model can be used to analyze the nature of hard instances and evaluate the performance of CSP algorit hms, and hence it helps the researchers to design more efficient algorithms.

Abstract:Resource reservation is an important means to guarantee the QoS of distributed m ultimedia applications. In this paper, an application level QoS negotiation prot ocol is proposed. It has two merits. One is globalizing the local QoS parameters so that the reservation becomes more flexible and users can reserve resources u sing the parameters they concern. The other is localizing the global reservation to make reservation simpler and to facilitate the implementation of optimal del ay allocation.

Abstract:This paper adapts Bordat's lattice construction algorithm to integrated mining, presents a fast algorithm to extract both association and classification rul es from concept lattice. It is shown that classification and association rules m ining can be unified under concept lattice framework. Experimental evaluation pr oved the validity of the algorithm.

Abstract:The performance of parallel programs, especia lly the parallel efficiency evaluation, usually needs to get data after running the program. Then through checking the experimental results, the programmer opti mizes the parallel algorithms, modifies data placement policies, or even selects other algorithms. In this paper, based on the BSP parallel programming model, a n evaluation model for predicting the efficiency of parallel programs is present ed, which allows programmer to evaluate the efficiency at the stages of algorith m design and programming, and then take some measures to optimize the programs f urther. The experimental results show that the evaluation predication model is a ccurate.

Abstract:Some problems plaguing the current researches on social norms for agent are pointed out in this paper. By importing Coleman' s foundations of social theory, a social norm conception based on rights allocat ion is proposed. The assumption of average rights allocation in Coleman's theory is relaxed, a formal definition for social norms in multi-agent system is give n, and the design of a norm as a constraint satisfaction problem is derived. The n, the rationality of social norms is investigated in this paper, and the condit ions for social norm's constitution and abolition are given. Finally, the ration ality property of social norms under symmetrical interest situation is investiga ted and a value equalization algorithm that can improve the rationality property of a norm is given. The work may contribute to the researches on market-orient ed multi-agent system.

Abstract:In this paper, a trainable and fast partial parsing method for Chinese is presen ted. This method is based on the augmented cascaded finite state automaton propo sed in this paper, which uses error-driven machine learning to achieve high acc uracy. The preliminary result shows that the proposed method is promising.

Abstract:In this paper, an evolutionary development mo del MHSC/DM (methodology for high-level specification construction) is proposed based on MHSC, which supports the executable specification, and the composing r oles and their interrelations are illustrated. This paper addresses the transfor mation classification, system generation and configuration, and architecture of the model. The development model can be used to support automatic generation of prototype system quite well from requirements through evolution and can keep sys tem consistent.

Abstract:The wide use of database and the appearance o f data warehouse technology provide new direction for Decision Support System(DS S). Online Analytical Processing (OLAP) is one of these new DSS technologies bas ed on data warehouse. The mid-database design, multidimensional data analytical models and OLAP analytical system module structure for insurance business are p resented in this paper, and the implementation of OLAP function setup subsystem and main analytical subsystem are outlined. It meets the requirement of the 12 r ules of E.F. Codd, and makes improment based on practical application.

Abstract:Border Gateway Protocol (BGP) is a distribute d dynamic routing protocol for exchanging routing information among autonomous s ystems. In this paper, the implementation of distributed routing protocol BGP-4 in high performance router developed by Tsinghua University is presented. A met hod based on event-driven programming is presented to implement the complicated protocol. Event-driven programming can reflect the real situation of networks, so it is suitable for the network programming. This method can also be used to implement other higher layer protocols of Internet.

Abstract:It is an important subject to build the model of mental states for Agents in the theoretic research of agent. A model BDICU ( belief desire intention combined with utility) of mental states combined with ut ility is presented in this paper, which guarantees agent's action with rational logic and decision, the problem of side effect with a solution is provided, and the rules for producting and updating goal and intention are also presented. Thi s model modifies and extends Rao and Georgeff's Belief-Desire-Intention theory , and provides rational agent systems with implementing support on logic and utility.

Abstract:A hybrid method is presented for automatic Chinese discourse structure analysis in this paper. This method is based on RST (rhetorical structure theory) analysi s, thematic progression analysis and other linguistic methods. Vector space mode l and other statistical methods are also employed to inhance its robustness. A d eterministic RST analysis algorithm is proposed and implemented. Compared with o ther existing methods, the proposed method has better applicability and precision.

Abstract:Simplified and unsimplified (traditional) Chinese character conversion is very i mportant for the cultural and economic exchange between the mainland of China an d overseas Chinese speaking areas. In this paper, the design and implementation of an automatic conversion system is presented, which has been put into use. The system architecture is shown firstly, and some important data structures, the w ord dictionaries and mapping tables are presented. Then, the system processing f low is outlined, which includes preprocessing and post processing, ambiguity res olution, word conversion, etc. Finally, the system performance analysis and the test result are listed.

Abstract:The disadvantages of localization are pointed out in this paper, Java's intel encoding is also analyzed profoundly, and the c ompiler supporting internationalization is designed and implemented. With the ch aracter encoding transformation scheme, international characters and strings hav e the same operations as the English characters and strings, the storage spaces of class files are saved, and the applications have conditional encoding-indepe ndence and the problem of many Chinese character sets is solved to some extent. Separating compiler messages from codes independent of locale context supports t he implementation of international information. Small font libraries are adopted to implement the real platform-independence of applications including local ch aracters, the problem JKD (Java development kit) has not solved, and the charact ers' encoding transformation is performed during loading a class file. To solve the compatibility problem between different JDK versions, the different encoding transformation schemes are chosen according to the character's encoding charact eristic. The test results prove that the internationalization scheme designed in this paper can solve the internationalization problems thoroughly.

Abstract:The research on bilingual corpora is one of t he most important topics in the field of natural language processing. Paragraph alignment is to find the translation paragraph pairs between source text and tar get text. Paragraph alignment is a necessary and important phase between section alignment and sentence alignment, but it is disregarded at present. This paper aims at dealing with this process. Because it is difficult and unnecessary to al ign paragraphs directly, this problem is changed into a “segmentation and align ment” one. This paper finds some anchors for segmentation first, and then match es the sentences which contain these anchors, finally checks the validity of the results and segments the texts into translation blocks. The experimental result s show that the method can achieve good results and can also be applied to other language pairs.

Abstract:A number of papers concerning the off-line r ecognition of handwritten Chinese characters have been published in the recent y ears, and almost all of them focus on the recognition of isolated characters. Ho wever, off-line recognition of handwritten Chinese characters is not satisfacto ry. One reason is that the recognition rate is low, the other is that the segmen tation of handwritten Chinese characters is a difficult problem because recognit ion of characters relies on correct segmentation of characters. In this paper, a ccording to structural features of Chinese characters, elements are merged based on their topological relations, viz., upper-bottom, left-right and inside-ou tside. The width of elements and the spacing between neighboring elements in the whole handwritten Chinese character string are analyzed to guide the merging of left-right elements. Experimental results show that the method has satisfactor y performance for segmenting handwritten Chinese character string.

Abstract:The Gamma formalism is a high level parallel programming tool without artificial sequentiality. P. Fradet and D.Le Metayer pr oposed recently a model of structured Gamma which makes it easy for the programm er to structure the data and to specify particular control strategies. It also m akes the automatic analysis of programs easier. Without relying on the “address ” concept used by them, a more general form of the structured Gamma is proposed and its syntax and semantics are described in this paper. Types of structured m ultisets have been discussed at two levels. The author also shows that the new d efinition keeps the theoretic results, such as static type checking of programs untouched.