Abstract:An integrated semantic analysis system is presented, and the structured language models are proposed based on it. The semantic analysis system can automatically tag semantic class for each word and analyze the semantic dependency structure between words with the precision of 90.85% and 75.84% respectively. In order to describe sentence structure and long-distance dependency, two kinds of structured language models are examined and analyzed. Finally, these two language models are evaluated on the task of Chinese speech recognition. Experiments show that the best semantic structured language model?headword trigram model?achieves 0.8% absolute error reduction and 8% relative error reduction over the trigram model.

Abstract:An approach is presented to extract roads from aerial city images based on the Dempster-Shafer evidence theory. A road model is a priori constructed. Aerial images are divided into sub-blocks from which regions consisting of 8-connected pixels with similar gray scales are obtained, and regions with relatively big areas are selected as candidate road segments. Then Dempster rule is applied to compute the fusion of basic probability assignment functions (BPAF) defined respectively on the features extracted from the candidate road segments and on the original road model. Finally, the BPAF fusion is utilized to find the conclusive road segments and these road segments are connected and pruned to form the de facto roads. Experimental results demonstrate the ability of the D-S evidence theory based approach to accurately extract roads from aerial city images.

Abstract:Classic Mean-Shift based tracking algorithm uses fixed kernel-bandwidth, which limits the performance when the object scale exceeds the size of the tracking window. Based on the analysis of similarity of object kernel-histogram in different scales, i.e. the Bhattacharyya coefficient, a theorem is found and proved i.e. the changes of object scale and position within the kernel will not impact localization accuracy of Mean-Shift based tracking algorithm. Using this theorem an automatic bandwidth selection method is proposed based on backward tracking and object centroid registration. The proposed method is applied to track vehicle changing in size with encouraging results.

Abstract:Cell’s immunological information is acquired through analyzing color immunofluorescence image. A pixel growing algorithm with dynamic seed values is proposed to extract fluorescent regions in the CIE (L*, a*, b*) uniform color space. Ellipse matching method is proposed to extract cells from the obtained fluorescent regions. Finally antigen information is acquired through computing and decomposing color of the cell with the pre-computation and discretization method. The methods proposed for the acquirement of cell’s immmunological information are appropriately evaluated. The experiment indicates that acquired the immunological information meets practical medicine analysis.

Abstract:Determination of the hierarchical relationship and the objective patterns of sub-classifiers is a primary problem in the construction of a hierarchical classifier. In this paper, a method focusing on the similarities between patterns is proposed to generate a hierarchical structure automatically. Firstly, a similarity measurement utilizing the confusion matrix is advanced to avoid the drawbacks of the traditional measurements, such as high computation costs and invalidity of preliminary conditions. Then abiding by Fisher’s Principle, a Patterns’ Similarity Relationship Analyzing Machine (PSRAM), which is integrated with the supervised and unsupervised pattern recombination methods, is designed to adaptively construct the structure of a hierarchical classifier. Various tests are testified that the proposed method is effective and practical, and it can prominently improve the performance and robustness of the hierarchical classifier.

Abstract:The background subtraction is an important method to detect the moving objects, and its difficulty is the background update. So a background reconstruction algorithm based on pixel intensity classification is presented in this paper. According to the hypothesis that the background pixel intensity appears in image sequence with maximum probability, the pixel intensity differences between sequential two frames are calculated, and the intensity values at the pixels are classified by means of these differences. For the new algorithm, neither the pre-training of the background without any moving target, nor the models of background and targets are needed. Simulation results indicate that background can be reconstructed correctly by using the new algorithm, so the target can be extracted perfectly and tracked successfully.

Abstract:Classification is important in data mining and machine learning. In this paper, a classification approach based on evolutionary neural networks (CABEN) is presented, which establishes classifiers by a group of three-layer feed-forward neural networks. The neural networks are trained by an improving algorithm synthesizing modified Evolutionary Strategy and Levenberg-Marquardt optimization method. The class label of the identifying data can first be evaluated by each neural network, and the final classification result is obtained according to the absolute-majority-voting rule. Experimental results show that the algorithm CABEN is effective for the classification, and has the better performance in classification precision, stability and fault-tolerance comparing with the traditional neural network methods, Bayesian classifiers and decision trees, especially for the complex classification problems with many classes.

Abstract:Recently, a new manifold learning algorithm, LTSA (local tangent space alignment), has been proposed. It is efficient for many nonlinear dimension reduction problems but unfit for large data sets and newcome data. In this paper, an improved algorithm called partitional local tangent space alignment (PLTSA) is presented, which is based on VQPCA (vector quantization principal component analysis) and LTSA. In the algorithm, the sample space is first divided into overlapping blocks using the X-Means algorithm. Then each block is projected to its local tangent space to get local low-dimensional coordinates of the points in it. At last, the global low-dimensional embedded manifold is obtained by local affine transformations. PLTSA is better than VQPCA in that it gives the global coordinates of the data. It works on a much smaller optimization matrix than that of LTSA and leads to a better-scaled algorithm. The algorithm also provides a set of transformations that allow to calculate the global embedded coordinates of the newcome data. Experiments illustrate the validity of this algorithm.

Abstract:Clustering algorithms are wildly used in the research of microarray data to extract groups of genes or samples that are tightly coexpressed. In most of them, some parameters should be predefined artificially, however, it is very difficult to determine them manually without prior domain knowledge. To handle this problem, an iterative clustering algorithm is proposed. Firstly, by sorting the original data by dominant set, similar genes would be aligned together. It’s hard to specify the cluster boundary. A criterion is presented to partition a cluster from the sorted data according to the property that the distances between the inside elements are smaller than that of outside elements. The idea is to remove the cluster form the current data set, repeat the process, and stop the algorithm when the stop criterions are satisfied. The new clustering algorithm is analyzed on several aspects and tested on the published yeast cell-cycle microarray data. The results of the application confirm that the method is very applicable, efficient and has good ability to resist noise.

Abstract:This paper mainly fills the gap of diagnosis test for non-observable common variables or high cost of testing common variables. The concept of replacement test is presented as a new alternative to diagnosis test. The concept of relevant replacement test is proposed for some observations, and makes best use of the effects of component replacement upon the observations of the system being diagnosed to characterize the discrimination of candidate diagnoses, and the generation of new conflicts. Based on it, the concept of replacing decomposition for diagnostic problems is proposed by virtue of the characteristics of replacement, and the decomposition of the system being diagnosed through replacing the components in the intersection set of some subsystems with normal components directly is investigated. In non-observable common variables or high cost of testing common variables, the results in this paper can improve the adaptation and the effectiveness of diagnosis tests, reduce the cost of the test, and provide a theoretical basis for decomposing the system being diagnosed.

Abstract:As the detection key in symmetric watermarking scheme can be used to forge or remove watermarks from digital works, it is required that the detection key be secret in watermark detection procedures. Based on zero-knowledge and proof of knowledge concepts and protocols in Cryptology, zero-knowledge watermark detection protocols can make the verifier believe the presence of a watermark in a disputed digital work while not compromising the detection key. The security requirements of a publicly verifiable zero-knowledge watermark detection scheme are outlined in this paper. Then a publicly verifiable commitment scheme and a zero-knowledge proof of knowledge protocol which proves knowing the discrete logarithm of a committed value are presented. Finally, using the above scheme and protocol as building blocks, a publicly verifiable zero-knowledge watermark detection protocol is proposed and its security considerations are addressed.

Abstract:In the rapidly expanding mobile environment, authenticity of communicating parties is one of the big research challenges and is receiving increasing attention. In the Mobile IPv6 defined by IETF (Internet engineering task force), IPSec (IP security) protocols and RR (return routability) mechanism are used to protect signaling between related communicating nodes, however, how to realize identity authentication has not been efficiently solved. In this paper, the advantages and disadvantages of two authentication techniques?certificate-based authentication and identity-based authentication are analyzed. The scalability of certificate-based means is excellent, but the deployment of PKI (public key infrastructure) and the distribution of certificates make this method costly. On the contrary, identity-based method hurdles the deficiency of certificate-based means, nevertheless the scalability suffers from the share of parameters among related nodes. Then an approach of integrating the two methods mentioned above is proposed to realize a secure and fast authentication with low cost and high scalability. Finally, this hybrid technique is applied in Mobile IPv6 to improve the negotiation of SA (security association), and the security issues are discussed.

Abstract:The existing publish/subscribe systems can’t match events with subscriptions based on the semantic of events, and they cannot support events with complex structure (such as graph structure). The Semantic Web technologies are introduced into the publish/subscribe system and an ontology-based publish/subscribe system is proposed. In this system, the concept model of events is represented as ontologies, the events are represented as RDF graphs, and the subscriptions are represented as graph patterns. The system can overcome the disadvantages of the existing publish/subscribe systems. Experimental results show that it has high matching efficiency.

Abstract:As an effective supplement to end-to-end congestion control mechanism, active queue management aims to keep high link utility while maintaining low queuing delay. As an effective mechanism, the FIPD (fuzzy intelligent packet dropping) algorithm provides a brand-new method for active queue management. However, the FIPD algorithm also has its intrinsic defects, such as high packet loss rate. The objective of this paper is to overcome the defects of FIPD, and a new active queue management method FIPE(FIPD with ECN) is also proposed.Firstly, in this paper, the algorithm FIPD is reviewed and analyzed. Then the advantages and disadvantages are also pointed out. In order to overcome the defects such as high packet loss rate, the well-known ECN mechanism is introduced, which greatly improves the successful packet transmission. At the same time, some amendments are employed and a new active queue management algorithm FIPE is developed. Finally, the validity of the new algorithm is verified by a series of simulations on NS2 simulator.

Abstract:Manycast is a new "one-to-some-of-many" communication pattern. It is the general communication form of anycast and multicast. With manycast a client can contact several servers to increase reliability or a distributed service can disseminate the important information to more than one server throughout a MANET network. This paper describes the main issues with manycast and mechanisms proposed in the last a few years. It also discusses some possible optimizations by taking advantage of other related work and shows some applications that have already been deployed in practice.

Abstract:The characteristic that nodes can enlist into the network topology freely and independently without any fixed infrastructure makes mobile Ad hoc networks (MANET) widely used in various environments such as disaster rescue, battlefield and so on. Conventional mobile Ad hoc routing protocols usually concentrate on the constrained condition of ‘shortest path’ with minimum hops measurement. However, related researches show that the path with minimum hops can’t provide the minimum end to end delay guarantee. Moreover, recently, Ad hoc network is required to support the delay-sensitive traffic. So the reduction of the end to end delay is a new challenge for Ad hoc networks. To this point, this paper mainly focuses on the node delay and a cross-layer method is used to predict the end-to-end delay. Finally, a new routing protocol Delay Oriented Adaptive Routing (DOAR) is presented, which is based on a ‘minimum prediction of delay’ mechanism. Simulation results show that the derived path length in the proposed DOAR protocol is slightly higher than that of Dynamic Source Routing (DSR) protocol, but it can significantly reduce the average end-to-end delay in both static and mobile scenarios.

Abstract:QoS routing based on node-delay information is an active research area these years. When delay is used as node state, it is often assumed that the state between each node is independent. Node-State independence assumption can provide a more tractable solution to delay constrained routing, especially when state information is Probability Density Function (PDF) of delay at each node. In this paper, the effectiveness of node-state independence assumption in Autonomous Systems with limited node number is investigated, and the conclusion is verified through vast simulation, whereas 60 link-delay PDFs and 15 path-delay PDFs in the network are observed. The 15 path-delay PDFs are also calculated by convolving the PDFs of contributing link-delays based on independence assumption. The statistical distance between the two sets of path-delay PDFs is measured by calculating divergence and comparing their delay expectation, variation and loss probability. Simulation and analysis results indicate that for Autonomous Systems with limited node number, node-state independence assumption is reasonable and path metrics calculated based on this assumption approximates well the original values and can be used in QoS routing.

Abstract:Traditional blind additive SS (spread spectrum) watermarking systems have such features: high robustness, low detection value and small capacity. The main reason leading to low detection value is the host signal’s interference. Based on the Gaussian signal detection principle under Gaussian noise, the correlation detection is analyzed in theory. In addition, the using condition of correlation detector and the relationship between the normalized correlation detector and SNR (signal noise ratio) are discussed in this paper. Based on these analyses, an improved blind additive watermarking algorithm is designed. Experimental results show that the algorithm effectively holds back the interference of host signal and the performance of detection has been greatly improved. Furthermore, it has greater robustness and imperceptibility.

Abstract:The measurement studies show that the burstiness of packet traffic in LAN as well as WAN is associated with self-similar and long-range dependency, and Hurst index is the key value of this model representing the burstiness of traffic. With the analysis in discrete wavelet domain, the nature of the wavelet coefficients and their statistical properties are proposed. Then an adaptive, efficient unbiased estimator of Hurst index based on multiresolution wavelet analysis and weighted regression is presented. Simulation results based on fractal Gaussian noise and real traffic data reveal the proposed approach shows more adaptiveness, accuracy and robustness than traditional estimators which has only O(N) computation. Thus this estimator can be applied to the application of traffic management and real-time control in high-speed networks.