Search

This paper presents the design of microwave bandpass filters with frequency-dependent (FDC) coupling for power wireless private network communication. The filters are composed of four resonators arranged in the in-line and folded topology using substrate integrated waveguide (SIW) technology. The frequency-dependent coupling is implemented bya magnetic coupling iris and a balanced microstrip line with a pair of metallized via-holes, which produces an extra zero in comparison to classical approaches. To verify the characteristics of the proposed FDC, an in-line fourth-order substrate integrated waveguide bandpassfilter (BPF) with one transmission zero and a folded configuration BPF with one small negative slope for independently controlled transmission zero are designed.

The method of symmetrical spanwise blowing was used to suppress the wing-body rock of a typical canard-configuration model in free-to-roll tests. The experimental results showed that the rock was successfully suppressed by symmetrical spanwise blowing when the pitch angle was between thirty degrees and thirty-four degrees. However, this kind of blowing failed to suppress the non-commanded motion once pitch angle exceeded thirty-four degrees. To analyze the mechanism under this phenomenon, surface-pressure measurements and PIV tests were carried out. The reason for the successful suppression was inferred as the success of the delaying of windward vortex breakdown at a small sideslip angle, while the failed results at larger pitch angles were due to the fact that the blowing cannot delay the windward breakdown anymore under very small sideslip.

In this paper, we investigate the popular neural word embedding method word2vec as a source of evidence in short text similarity calculation, which has been applied to intention classification of user queries in our intelligent customer service system — Alime. For avoiding information loss on sentence-level vector representation, we design a Word Similarity Maximization (WSM) method, which utilize word-level embedding vectors to directly calculate query similarity. Experimental results on a dataset from real daily user query data from Alime system show WSM can better detect query-intention relation than baseline methods.

Word sense disambiguation (WSD) plays an important role in natural language processing, such as information retrieval, text summarization, question answering, machine translation, etc. Abbreviation disambiguation is an important sub-domain of WSD, and in this paper we try to construct a co-occurrence knowledge base for realizing fast abbreviation disambiguation, since speed is essential in applications such as question answering and machine translation. In our proposed method, named entities in context of abbreviations are essential for this disambiguation task, and the final experimental results show that our method can effectively realize the mentioned aim.

Fast diffusion of information makes microblogging an efficient platform for seeking information, and the most convenient way for a user to get useful information is following others and in real time the updates of the followed users will be presented automatically. Deciding to follow or not to follow a user is usually based on this user's tweet content, because users usually try to find friends who post information relevant to their interests. Since reading through all tweets of a user is very time-consuming, we need to design a convenient user summarization form. In this paper, we design a model for summarizing users with their most representative tweets. The experimental results on Sina-weibo, one of the most popular microblogging sites in China, show our model can get a better performance than baseline methods.

An array PSJA is tested in this paper by using high-speed phase-lock schlieren system. The characteristic of the transient flow field is studied. The experiment shows that a series of high-speed blast waves spreads outward after discharge. These blast waves, attenuated to the local sound speed, form a wavefront, which continues to influence the flow field. Due to the interaction between each blast wave and the squeeze from previous blast waves, the subsequent blast waves' impact range is limited and form “jet-like” phenomenon. A little later, there comes a high-speed jet to continuous influence the flow field.

As one of the most important methods of vulnerability mining, the technology of fuzzing is developing rapidly and widely used during vulnerability analysis of various kinds of software. Nowadays, with the widespread craze of Android phones, the security of Android operating system has drawn more and more attention. However, fuzzing technology of Android is still in the beginning and professional fuzzing products of Android are unavailable. In this article, a fuzzing method of Android browser based on bitmap structure is presented. The result of experiment proves that this method is efficient in triggering Android browser crashes and can help with further work of analysis and exploitation.

Symbolic execution simulates program execution by replacing concrete values with symbolic variables for inputs. It could be used in software behavior analysis, vulnerability detection and software security assessment. In this paper, we analyze the path explosion problem encountered in vulnerability detection with the state-of-the-art symbolic execution technology for large scale file parsing programs. We also propose 4 alleviations to ease the problem, i.e. loop controlling, irrelevant path elimination, path selecting and parallel symbolic execution. Based on these alleviations, we implemented a prototype tool to detect file parsing vulnerability in large scale programs automatically, and evaluate it with a suit of benchmarks chosen from open source programs. Our tool detected not only all reported vulnerabilities of memory overflow in the benchmarks, but also some unreported vulnerabilities. The evaluation results show these alleviations could effectively ease the path explosion problem while analyzing large scale file parsing programs.

To meet different design requirements and improve the efficiency of antenna design, aggressive space mapping (ASM) algorithm combined with finite element method (FEM) is used to design a microstrip antenna for the first time. We exploit a coarse-mesh and a fine-mesh of FEM as the coarse model and the fine model respectively. The optimization results show the effectiveness of the method. It needs less calculation time and increases the design efficiency comparing with only optimizing the fine model.

A novel fourth wavelength Folded broadband microstrip antenna is presented in this paper. This patch antenna is used half structure technique and folded patch technique to achieve a wide impedance broadband patch antenna with small geometric size. It is employed the same parameters as the folded U-slot quarter-wave patch except with half the area and it is shown that the impedance bandwidths, radiation patterns, radiation efficiencies and gains of the half structures are comparable to the corresponding full structures.

A novel compact broadband multilayer microstrip antenna with half u-slot is presented in this paper. This patch antenna has two layers, the top layer and the middle layer. Both layers of the patch antenna were connected to the ground plane with shorting wall. In this paper, we proposed a multilayer with half u-slot technique to achieve a wide impedance broadband patch antenna with small geometric size. The VSWR, gain and radiation pattern of this antenna are simulated. The measured VSWR of it is also shown in the paper. The results are discussed in detail

A novel compact broadband multilayer microstrip antenna is presented in this paper. The VSWR, gain and radiation pattern of this antenna are simulated. The measured VSWR of it is also shown in the paper. The results are discussed in detail.

This paper proposes a strategy for retrieving multi-texture images based on contour and texture segmentation. Firstly, the contour of each texture primitive is extracted from an image and its Fourier descriptor is calculated. Thus, the contours of the texture primitives in the original image are clustered according to these shape descriptors. Then Gabor wavelet transform is applied to extract the features of texture primitives for each group, so the image can be represented by a set of feature vectors in feature space. Finally, an improved and noise insensitive Hausdorff distance is used to calculate the distance between two feature vector sets. Furthermore, the retrieval of multi-texture images can be implemented. A large amount of experiments prove that our method has higher retrieval precision, compared with the state-of-arts methods.

Tikhonov-type regularization method for noisy chaotic time series prediction is investigated. The current regularized local prediction method is interpreted as one kind of filter factors to decrease the variance of the predictor. One drawback in the interpretation is the ignorance of the random noise in coefficient matrix, another drawback is the relationship between the regularization parameter and the noise condition is not clearly explained, so the determination of regularization parameter has to resort to some techniques such as cross validation. In this study, local linear model is studied from the perceptive of the errors-in-variables (EIV) modeling, and the predictor is designed by considering the noise both in coefficient matrix and right-hand side. The optimal solution can be obtained by second order convex program (SOCP) if given a perturbation bound of the noise, and the solution can be reformulated as a form of Tikhonov regularization, and it will be shown how regularization parameter is related to the Frobenius norm of the noise containing in coefficient matrix and right-hand side. Two demonstrations are presented to show the validity of the results.

Multivariate time series are common in experimental and real systems. According to the embedding theory, in the absence of observational noise only one time series should be needed to recover dynamics. However, for real data, the noise always exist. There may be large advantages in using more measurements. In this paper, we focus on the issue of using multivariate time series to model and predict. The experiments show that by using multivariate time series the influence of noise could be reduced. Since the structure of the embedded time series is complex, the singular value decomposition (SVD) is used to extract feature components in the multivariate time series. Then the neural network (NN) is applied for identification of the dynamic system. The effectiveness of this method is shown by simulation of the real world multivariate time series as well as a well-known chaotic benchmark system.

High-dimensional index is one of the most challenging tasks for content-based video retrieval. Typically, in video database, there exist two kinds of clues for query: visual features and semantic classes. In this paper, we modeled the relationship between semantic classes and visual feature distributions of data set with the Gaussian mixture model, and proposed a semantics supervised cluster based index (briefly as SSCI) approach to integrate the advantages of both semantic classes and visual features. The entire data set is divided hierarchically by a modified clustering technique into many clusters until the objects within a cluster are not only close in the visual feature space but also within the same semantic class, and then an index entry including semantic clue and visual feature clue is built for each cluster. Especially, the visual feature vectors in a cluster are organized adjacently in disk. So the SSCI-based nearest-neighbor search can be divided into two phases: the first phase computes the distances between the query example and each cluster index and returns the clusters with the smallest distance, here namely candidate clusters; then the second phase retrieves the original feature vectors within the candidate clusters to gain the approximate nearest neighbors. Our experiments showed that for approximate searching the SSCI-based approach was faster than VA+-based approach; moreover, the quality of the result set was better than that of the sequential search in terms of semantics.

Traditionally, how to bridge the gap between low-level visual features and high-level semantic concepts has been a tough task for researchers. In this article, we propose a novel plausible model, namely cellular Bayesian networks (CBNs), to model the process of visual perception. The new model takes advantage of both the low-level visual features, such as colors, textures, and shapes, of target objects and the interrelationship between the known objects, and integrates them into a Bayesian framework, which possesses both firm theoretical foundation and wide practical applications. The novel model successfully overcomes some weakness of traditional Bayesian Network (BN), which prohibits BN being applied to large-scale cognitive problem. The experimental simulation also demonstrates that the CBNs model outperforms purely Bottom-up strategy 6% or more in the task of shape recognition. Finally, although the CBNs model is designed for visual perception, it has great potential to be applied to other areas as well.

Linear classifiers are of great significance in the classification field. In this paper, with meticulous studies on the linear classifiers, we gained a general framework for constructing fast trained linear classifiers, and found the key point that dominated the performance of the classifiers. Based on that result, we brought forward a mutual information based (MI-B) linear classifier. Experiments showed that the new method achieved a striking performance improvement over all the existing fast trained linear classifiers and worked stably. In some dataset it even outperformed the linear support vector machine (LSVM)

For chaotic system prediction, ESNs (echo state networks) are realization of neural state reconstruction, in which the reconstructed state variable is from the internal neurons' activation, rather than the delay vector obtained from delay coordinate reconstruction. In the framework of the neural state reconstruction, some quantitative analyses can be further made on the issues such as the network structure configuration and initial state determination. Based on the simulation study on chaotic data from Chua's circuit, it is shown that the ESN is a non-minimum state space realization of the target time series, and the initial state can be freely chosen in the training process, and in the phase of prediction, ESN needs to know where the prediction begins by being set a proper initial state through a process of teacher forcing.

Because the chaotic system is initial condition sensitive, it is difficult to decide a proper initial state for a recurrent neural network to model observed one-dimension chaotic time series. In this paper, a recurrent neural network with feedback composed of internal state is introduced to model one-dimension chaotic time series. The neural network output is a nonlinear combination of the internal state variable. To successfully model a chaotic time series, this paper proves that the recurrent neural network with internal state can start from arbitrary initial state. In the simulation, the neural systems perform multi-step ahead prediction, also, the reconstructed neural state space is compared with the original state space, and largest LEs (Lyapunov exponents) of the two systems are calculated and compared to see if the two systems have similar chaotic invariant.