Abstract:Signer-Independent sign language recognition is an unavoidable problem that must be solved in order to promote the practicality of sign language systems. In the signer-independent sign language recognition research, the lack of training data and the signer-independent sign language data variation bring a challenge to the effectivity of the existent research frame. This paper proposes a new research frame for signer-independent sign language recognition, and provides the strategy and ideas to solve the problem. Finding resolution to these problems is significant not only to the research on Chinese sign language recognition but also to other related fields.

Abstract:In this paper, a similarity relation between two predicates is defined first. To a given predicate, the set of action for the predicate can be obtained by the similarity relation. Then, the domain knowledge is extracted from the common fluent in preconditions and effects of all actions for each set of action, and the formalism for the domain knowledge is given. Finally, the contradictions in the initial states and the goal states in a particular planning problem with domain knowledge can be discovered. The strategy of extracting the domain knowledge is integrated in the planner StepByStep, and the domain knowledge is the necessary theory when one predicate by action is realized in the planner.

Abstract:To solve the time-consuming problem of the fitness assignment in the multi-objective evolutionary algorithm, this paper proposes a novel fitness assignment—dominating tree. The dominating tree preserves the necessary relationships among individuals, contains the density information implicitly, and reduces the comparisons among individuals distinctly. In addition, a smart eliminating strategy based on the dominating tree maintains the diversity of the population without extra expenses. A new multi-objective evolutionary algorithm based on dominating tree is proposed on these innovations. By examining three performance metrics on six test problems, the new algorithm is found to be competitive with SPEA2 and NSGA-II in terms of converging to the true Pareto front and maintaining the diversity of the population, moreover, it is much faster than other two algorithms.

Abstract:A synthetic fingerprint generation method is presented and implemented in this paper. A new combination model for the orientation field of fingerprint is applied to generate more realistic fingerprint direction map, and a new expression of the fingerprint density map is presented. Then a modified Gabor filter is used for ridge pattern generation. This method consists of two main processes: First, a master-fingerprint image is generated after direction map generation, density map generation and ridge pattern generation. Then a realistic synthetic fingerprint is obtained through a series of steps on the master-fingerprint, including adding scratches, displacement of ridges, erosion or dilation of ridges, distortion, noising and rendering, global displacement and rotation, changing contrast and adding background noise. The synthetic fingerprint generation platform implemented with this method has been applied and verified in Biometrics Verification Competition 2004. Finally, the numerical experimental results approve the validity and robustness of the method.

Abstract:A simple and efficient method based on semi-supervised learning technique is proposed for behavior modeling and abnormality detection. The method is composed of the following steps: (1) Dynamic time warping (DTW) based spectral clustering method is used to obtain a small set of samples to initialize the hidden Markov models (HMMs) of normal behaviors; (2) The HMMs’ parameters are further trained by the method of iterative learning from a large data set; (3) Maximum a posteriori (MAP) adaptation technique is used to estimate the HMMs’ parameters of abnormal behaviors from those of normal behaviors; (4) The topological structure of HMM is finally constructed to detect abnormal behaviors. The main characteristic of the proposed method is that it can automatically select the number of normal behavior patterns and samples from the training dataset to build normal behavior models and can effectively avoid the running risk of over-fitting when the HMMs of abnormal behaviors are learned from sparse data. Experimental results demonstrate the effectiveness of the proposed method in comparison with other related works in the literature.

Abstract:This paper introduces an unsupervised learning framework of Chinese syntactic structure based sentences similarity. First, all sentence pairs in the Chinese sentence corpus are aligned, and each pair is partitioned into similarity segmentations and different ones which alternately occur, Then, aligned similarity segmentations or different ones are selected as potential constituent candidates based on the strategy of similarity priority or of difference priority respectively. As the boundary friction may be introduced in the later step, its disambiguation is further carried out. Finally, by inducing sentence constituents, the syntactic structures are learned. In order to reduce word sparseness in the process, some words are replaced by classes in advance. Three forms of the sentence units, such as the sequence of words, the sequence of POS (part of speech)-tags and the sequence of words with POS-tag, are examined and the learned syntactic structures are evaluated respectively. The results show that different priority strategy achieves a better performance than the similarity one, and the Fs are above 46% for all three forms, with the best one being 49.52%, which is better than those having been reported.

Abstract:Summarization of video can well represent the content of video. In the MPEG-1/2 standard key frame is used for representing the content of a video sequence. Similarly, in an object-based framework of the suggested MPEG-4 standard, key video object can summarize the content of video objects. In this paper, an interactive key video objects selection model (IKVOS) is presented as the result of improving the model of key video objects selection (KVOS). The model of KVOS is proved to satisfy the criterion of induction. Also, the course of IKVOS is formalized and the model of IKVOS is proved to satisfy the criterion of coinduction. It can dynamically generate key video objects with user’s preference. The experimental results are given, as well as the evaluation of the performance of the proposed method whose distortion rate is lower than the current methods.

Abstract:This paper proposes a new domain-independent statistical model. In this model, four multiple discriminant analysis (MDA) criterion functions are defined and used to achieve global optimization in finding the best segmentation by means of the smallest within-segment distance, the largest between-segment distance and segment length. To alleviate the high computational complexity problem introduced by the new model, genetic algorithms (GAs) are used. Comparative experimental results show that the methods based on MDA criterion functions have achieved higher P_μ than that of TextTiling and Dotplotting algorithms.

Abstract:Semantic role labeling is a feasible proposal to shallow semantic parsing. A maximum entropy classifier is used in the semantic role labeling system, which takes syntactic constituents as the labeled units. Some useful features and their combinations are used in the classifier. In the post-processing step, only the roles with the highest probability among the embedding ones are kept. After predicting all the arguments, which have matched the constituents in full parsing trees, a simple rule-based post-processing is applied to correct the arguments that have not matched the constituents in these trees. F1=75.49% and F1=75.60% results are obtained on the development and test set respectively. So far as it is known, this is the best result based on single syntactic parser in literatures. Finally, some proposals for soving the difficulties in semantic role labeling and the future works are given.

Abstract:To solve the number of coalition structure increase rapidly, algorithm SCS (search of coalition structure), fast dynamic formation of Agent coalition, is given. It can prune the graph of Agent coalition structure, decrease the searching space., and proved that after pruning, the number of coalition structure is n(k-1)^n-k of that before pruning. Finally, an experiment is given. This work can be seen as an improvement of Jennings and Sandholm’s related work.

Abstract:How to implement software Agent is a key problem to develop Agent-oriented programming languages and development tools. Aiming to solve this problem based on the mainstream OO (object oriented) technology, this paper first discusses the differences between Agent and object, and then presents an implemental architecture and behaviors decision algorithm of software Agent. The architecture and related algorithm are based on OO technology and the simplified improvement of BDI (belief desire intention) Agent model. An OO-based design framework of software Agent is also proposed by using the POAD (pattern-oriented analysis and design) method. The approach is helpful to clarify how to extend the OO method suitably for developing Agent-oriented programming languages and development tools.

Abstract:This paper proposes an algorithm based on curve evolution for unsupervised texture segmentation. A multidimensional feature space is achieved by using a Gabor filter bank to extract texture features. To avoid deforming contours directly in a vector-valued space, a Gaussian mixture model (GMM) is used to describe the statistical distribution of the space and get the boundary and region probabilities. Then a framework of geodesic active regions is applied based on them to get final results. In the end, the experimental results demonstrate that this method can obtain satisfied boundaries between different texture regions.

Abstract:Geometrical active contours (GAC) are used extensively in computer vision and image analysis, particularly to locate object boundaries. However, GAC-based segmentations have the drawbacks of long evolving time and boundary leaking. Because halting speed fields (HSF) in GAC models are typically not smooth enough in homogenous region, they are not able to make the active contour quickly move towards the desired object boundaries. On the other hand, the HSF don’t really vanish along the object boundaries, thus the curve propagating can not stop on the object boundaries and continuously move into the object boundaries (boundary leaking). In this paper, an anisotropic diffusion model is therefore presented and then applied to the HSF in GAC model. This GAC-based segmentation with the diffused HSF can overcome the two drawbacks above. Experimental results on a synthetic image and two real world images show the improvements in terms of reducing both the segmentation time and boundary leaking, in comparison of GAC model with the original HSF.

Abstract:A robust bootstrapping framework, which employs Multi-EigenSpace modeling technique based on regression class (RCpMES) to build speaker models with sparse data, and a short-segments clustering to prevent the too short segments from influencing bootstrapping, are proposed in this paper. For a real discussion archived with a total duration of 8 hours, the significant robustness of the proposed method is demonstrated, which not only improves the speaker change detection performance but also outperforms the conventional bootstrapping methods, enen if the average bootstrapping segment duration is less than 5 seconds.

Abstract:Two strategies for information exchange between processors in parallel ant colony algorithm are presented. Theses strategies can make each processor choose other processors to communicate and to update the pheromone adaptively. A strategy is also presented to adjust the time interval of information exchange adaptively according to the distribution of the solutions so as to keep balance between the convergence speed and the diversity of the solutions. The adaptive parallel ant colony algorithm(APACA) based on these strategies adaptively updates the pheromone according to the equilibrium of the pheromone distribution in each information exchange so as to avoid the precocity and local convergence. These strategies are applied to the traveling salesman problem on the massive parallel processors(MPP) Dawn 2000. Experimental results show that the algorithm has higher convergence speed,speedup and efficiency than other parallel ant algorithms.

Abstract:Application-Layer multicast (ALM) is an important supplement to IP multicast. However, unlike in IP multicast, the participating nodes in ALM are selfish and strategic host users. In order to improve their own interests, selfish host users might not strictly obey the ALM protocols, because of which the overall performance of the multicast session could be affected. To design robust and trustworthy ALM protocols, it is necessary to study the selfishness in ALM. This paper surveys the recent research trends in this area, and classifies the researches into three categories according to the working steps of ALM protocols, that is, the selfishness in maintenance of control structure, the selfishness in collection of node information, and the selfishness in construction of data structure.

Abstract:Multi-constrained path (MCP) selection is one of the great challenges that QoS routing (QoSR) faces. Existing algorithms cannot make a good tradeoff among computation complexity, response speed and preventing from losing feasible solutions. Furthermore, neither linear path length function (LPLF) nor non-linear path length function (NLPLF) alone can solve QoS routing problems. A novel normal measure based path length function is defined and based on it, a normal measure based MCP (NMMCP) algorithm is proposed to solve m-constrained MCP problems. NMMCP makes a good tradeoff not only between on-demand computation and pre-computation, but also between LPLF and NLPLF based algorithms. By introducing Pareto optimal mechanism, NMMCP has nonlinear look-ahead ability. Extensive simulations show that NMMCP is very efficient when both performance and computation cost are taken into account.

Abstract:As a new Web pattern, Web service has been rapidly developed in recent years. How to dynamically integrate the existent Web services to form a newly value-added and complex service to meet the requirement of different users is a popular research area. This paper presents an algorithm GODSS (global optimal of dynamic Web services selection) to resolve dynamic Web services selection with QoS global optimal in Web services composition. The essence of the algorithm is that the problem of dynamic Web Service selection with QoS global optimal is transformed into a multi-objective services composition optimization with QoS constraints. The theory of intelligent optimization of multi-objective genetic algorithm is utilized to produce a set of optimal Pareto services composition process with constraint principle by means of optimizing various objective functions simultaneously. Theoretical analysis and experimental results indicate the feasibility and efficiency of this algorithm.

Abstract:A multiagents intermediary-driven combinatorial electronic market model is given. Applying multiagents coalition formation theory for it, a mathematical analysis methodology is used for customer coalition formation. At the macroscopic level of the marketplace, the combinatorial marketplace is described as a stochastic swarm system and the dynamics of multiagents coalition formation is studied. Based on the rate equation in swarm intelligence, a differential equation is derived, which gives the quantitative characteristic for coalition formation process of large-scale customers in the combinatorial market. Simulation experiments are performed to study the feasibility and validity of multiagents coalition formation in combinatorial trading. The global behavior of the system compared with correlative works proves the effective improvement of the dynamic process.

Abstract:Wireless sensor networks usually have limited energy and transmission capacity, and they can’t match the transmission of a large number of data. So, it is necessary to perform in-network compression or aggregation of the raw data sampled by sensors. By designing a ring topology, this paper proposes an algorithm for wavelet based spatio-temporal data compression in wireless sensor networks. The algorithm is capable of supporting a broad scope of wavelets that can simultaneously explore the spatial and temporal correlations among the sensory data. In this algorithm, the data in sensor networks are abstracted as a matrix, and the temporal and spatial correlation is then captured by the column and row wavelet transform respectively. The performance of the algorithm is qualitatively analyzed from the viewpoints of energy and delay. Theoretically and experimentally, it is concluded that the proposed algorithm can effectively explore the spatial and temporal correlation in the sensory data and provide a significant reduction in energy consumption and delay.

Abstract:Node mobility is one of the dominant factors causing decreased performance in mobile ad hoc networks and restricting network scalability. Selecting stable paths is an effective way to reduce the impact of node mobility. Current methods of selecting stable paths in mobile ad hoc networks suffer from several shortcomings. They may need a hardware function support for geographical position location or cross-layer function support for sending signal strength information to upper layers. In this paper, a new method of selecting stable paths based on the neighbor change ratio is proposed. Neither the special hardware support nor the cross-layer support is needed in this new method. NCR-AODV (neighbor change ratio Ad hoc on-demand distance vector) routing protocol is an extension of the AODV (Ad hoc on-demand distance vector) on-demand route protocol with the new method. The new protocol selects the path which has smaller hop counts and less local topology changes to forward data. The simulation results show that NCR-AODV protocol decreases the long path break probability and improves the network performance.

Abstract:In mobile ad hoc networks (MANET), node movement often makes the wireless link break, and results in the routing invalidation. In order to improve the stability of routing, based on the analysis of the time-t holding probability of single link, an on-demand routing protocol is provided based on the time-t holding probability of a whole path. With double reply of routing request from destination and reverse optimizing of middle nodes, this protocol chooses the path that has the maximal time-t holding probability as the best routing. Simulations show that, compared with AODV (ad hoc on-demand distance vector) and DSR (dynamic source routing), this protocol saves the overhead of the protocol and reduces the end-to-end delay of the packet.

Abstract:The topology of a wireless multi-hop network can be controlled by varying the transmission power at each node. The primary goal of topology control is to design power-efficient algorithms that maintain network connectivity and optimize performance metrics such as nodes lifetime and throughput. An energy-efficient distributed topology-control algorithm is proposed, briefly called VCGG (A varying-cone distributed topology-control algorithm on Gabriel graph). Using a new mechanism of selecting the logic neighbor nodes by firstly deleting the farthest node (FDFN), the VCGG algorithm builds a degree-bounded, power spanner and planar subgraph by making use of merit of the varying cone. The simulation results show that the VCGG algorithm performs better, in terms of power efficiency, number of communication neighbors and interference, than the existing SΘGG and SYaoGG algorithms.

Abstract:An application using a distributed hash table (DHT) with N nodes must choose a DHT protocol from the spectrum between O(1) lookup protocols and O(logN). However, various applications under different network churns require that an idea DHT would be adaptive in according with the churn rates. ROAD (routing on active and demand), a new lookup algorithm, adjusts itself to provide the best performance across a range of lookup delay and churn rates. The key challenges in the design of ROAD are the algorithms that construct the routing table’s size and decrease the delay. It will speed up the lookup process and reduce the service delay with the expressed routing table and power sorting multicast algorithm. Simulations show that ROAD maintains an efficient lookup delay versus churn rate tradeoff than the existing DHTs. ROAD should be expanded into a mechanism that provides some kinds of lookup services with a range of qualities of service through super-peers choosing methods.

Abstract:Braid group is a new considerable public key cryptography platform for the quantum computer ages, but almost all current intractable braid problems used for public key cryptosystems are flawy. The security of a braid public key cryptosystem can’t depend only on the hardness of conjugacy problems. By taking advantage of the non-conjugate transformation and multiple variant equations on braid groups, two intractable problems are proposed, and the hardness of these problems comes from the enlarged amount of variants. A new related public key algorithm and the analysis of its correctness, security, efficiency and parameter choice are subsequently presented. The new algorithm can resist current known attacks, and the ideal to combine some simple problems to a multiple variant difficult one is constructive for designing new public key algorithms.

Abstract:The trusted subject supports of the existing multi-level security models are reviewed and a new model called DLS (discrete label sequence) is proposed. It decomposes the lifecycle of a trusted subject into a sequence of untrusted states (US). Each untrusted state is associated with a certain current security label, and only the predefined trusted request events (TRE) can trigger the transition from one US to the other. Thus, the current security level of a trusted subject is dynamically changed according to its application’s logic. Definitions of secure states and rules to preserve security are also presented. Compared with the trusted subject implemented by security level range, this model gives a better support of least privilege and achieves the support within the MLS policy framework.

Abstract:An information disguising and hiding technique is used to protect DEM (digital elevation model) data for store and transformation in this paper. A method is designed to compress the DEM data with a very low bit ratio which would be hidden in the disguised data. The integer wavelet with parameter is used, and a wavelet coefficients set which can embed the information is built. This paper expands the just noticeable distortion (JND) analysis based on the human visual system (HVS) to apply it in the DEM data. Furthermore this paper constructs Hash one-way by the Rabin method, and the algorithm can protect two groups of the DEM data at a time and be public.

Abstract:This paper introduces a natural paradigm for fair exchange protocols,called ID-based partial proxy signature scheme. A security model with precise and formal definitions is presented,and an efficient and provably secure partial proxy signature scheme is proposed. This is a full ID-based optimistic fair exchange protocol. Unlike the vast majority of previously proposed protocols,this approach does not use any zero knowledge proofs,and thus avoids most of the costly computations.

Abstract:A new approach in optical burst switching(OBS) networks with QoS support is presented in this paper. In the approach,the data channels of an outgoing link at a core node are divided into multiple groups,with each group corresponding to a service class. The number of data channels in each group is mainly determined by data traffic. In general,a data burst(DB) can be sent on a data channel reserved by its burst head packet(BHP) only in its own group. Upon failing to reserve any bandwidth in its own group,the BHP tries to re-reserve even preempted bandwidth on a data channel in a lower-priority group. A lower-priority BHP can't reserve bandwidth on any data channel in a higher-priority group. In addition,the reasonable relation between the preempting DB length and the preempted DB length is also investigated in this paper.

Abstract:This paper proposes a fair bandwidth allocation mechanism FPIP(fair PIP). Dealing differently with the packets of long flows and short flows at routers,the mechanism can preferentially allocate the bandwidth of the router to short flows and allocate the remaining among the competing long flows. Furthermore,it can keep the queue length of the router at a reference value using a well-designed active queue management AQM(active queue management) algorithm. The simulation results show that this new mechanism can outperform CSFQ(core-stateless fair queueing) in terms of fairness,queue length and the response time of Web flows.