Earticle

How to extract the robust discrimination features is the key of face recognition (FR). Local binary pattern is one of the most widely used feature extracting method in FR for its comprehensive representation of the visual content of face image. However, the feature vector extracted by LBP is usually very high-dimensional and maybe contains information redundancy. To deal with the drawback of LBP, a novel nonlinear version of LBP is presented. The main idea is firstly all the feature vectors extracted by LBP are mapped into a feature space by a nonlinear mapping, and then the mapped features are expressed using the corresponding projection vectors. Lastly, FR is performed based on the projection vectors. Compared with LBP, the new method has two advantages. Firstly, it can capture the nonlinear information of the feature vector extracted by LBP. Secondly, it avoids the complex expression of the nonlinear mapping. The experimental results on two public standard visual face datasets demonstrate the proposed method is superior to LBP in recognition accuracy while its computational complexity is considerably reduced.

The research work in this paper is in the field, the moving target detection spatiotemporal correlation and difference contour tracking algorithm based on a fixed background. The algorithm in the background under the condition of fixed to pay a smaller time complexity, the target detection and tracking has a good effect, so it has higher application value. This paper mainly focuses on the study of motion estimation and compensation algorithm to eliminate the temporal redundancy. Algorithm of detection and tracking of video moving object is a core subject in computer vision field, but also the key technology of intelligent video surveillance system. It combines the research achievements of artificial intelligence and other fields of pattern recognition, image processing, has been widely used in every field of security monitoring, intelligent weapons, video conference, video retrieval. Therefore, detection and tracking algorithm research has the extremely important theory significance and practical value. The starting point of this article is the subjective quality of image reconstruction of how to improve the accuracy of motion estimation and compensation after, to reduce the computation complexity of motion estimation algorithms, to improve the efficiency of motion estimation. This paper makes some studies on the redundant wavelet domain block matching motion estimation and compensation, then the video image for non-translational motion, the DT triangular mesh motion estimation and compensation in the redundant wavelet domain to do related research.

With the development of artificial intelligence and pattern recognition technology, more and more research related to human face is constantly developing in all walks of life. At the present stage, the traditional face recognition algorithm based on LBP and SVM is not good, and the process of feature extraction and feature classification are deeply studied in this paper. For feature extraction, the authors put forward an improved CS-LBP texture feature; for feature classification, the author uses the histogram intersection (HIK) kernel function to classify the features which has high efficiency and good effect. Subsequently, experiments are carried out on the Yale data set and the ORL data set. Experimental results show that the proposed algorithm has a significant improvement on the face recognition effect of face direction change, and the illumination change is slightly improved. In the natural environment, most face recognition has the influence of human face direction and noise, and the effect of noise is a hot direction of face recognition research in the future.

We formulate model of kernel regression process which is exploited for image upsampling. The probabilistic process is studied for estimating missing information in an image. The term ‘linear regression’ is a designing tool for the relationship between a scalar dependent variable ‘b’ and one or more explanatory variables denoted ‘A.’ We provide some results of regression method that are tested on two natural images. Simulation results compare performance with various condition and parameter sets.

Because of the feature points can describe the local characteristics of the image in a reasonable manner, effective use of feature point of content based image retrieval become the current hot issues in the field of computer vision. Aiming at this problem, we put forward a kind of combination clustering based on feature points, a new method of image retrieval. The method includes the combination of feature point clustering algorithm and based on the algorithm of local color histogram construction strategy. With the existing and local color histogram retrieval method based on feature points, compared to the method can effectively solve the current method of feature point location information and feature point center relying too much on the problem. Subjectivity and as a result of the manual annotation image accuracy, the traditional image retrieval methods cannot meet the needs of the user. Multidimensional indexing technology is only from the perspective of how to improve the indexing algorithm to adapt to the large-scale database to consider a problem, in content-based image retrieval. Our research combines the advantages of the semantic analysis and kernel clustering which will enhance the performance of the traditional image retrieval methods and strengthen the feasibility of the algorithm.

The relationship between the nuclear parameters and model performance is complex, which is from relevance vector machine (RVM) regression model based on Gaussian radial basis kernel function. Aiming at the problem of how to determine the kernel parameters of RVM, a method to selecting kernel parameter of RVM based on AIC criterion is proposed. Firstly, a novel of statistic Q is proposed based on “Akaike” Information Criterion (AIC), while the Q is as a fitness function. Secondly, we use the differential evolution algorithm (Differential Evolution Algorithm, DE) to find the best kernel parameter, in order to choose determine the kernel parameters. Finally, a RVM regression model mode is established and it is used in predicting gold price. Experimental results show that the prediction model has higher precision and better fitting the generalization ability than the traditional method, which demonstrates the AIC-based criteria for selecting RVM kernel parameter method is effective and feasible.

Currently, with the increasing number of embedded applications, the research for the security of embedded systems has become popular. Considering the requirement of integrity and security, the characteristics and the design constraints of embedded systems, a method of constructing software trusted root and a trunk-branch trusted chain transfer model is proposed. The software trusted root is composed of boot loader and reliable kernel. SHA-1 engine which is included in boot loader (burned in boot flash), measures and loads the trusted kernel and protect the boot loader by prohibiting kernel and user-level application to write to flash. SHA-1 and trusted kernel together can be used as a trusted root to withstand non-physical attacks. Trusted kernel contains virtual trusted platform module (vTPM) module which can provide cryptographic functions and related services to the user kernel and guides flash to open up specific memory for platform configuration register (PCR) of vTPM. Application is running as a process of the trusted kernel. In the trunk-branch trusted chain transfer model, boot loader verify the trusted kernel, trusted kernel authenticate users kernel, and kernel users verify the application, thus the trust extents to the application layer. The proposed method not only avoids the trust attenuation problem which may occur in chain transfer, but also raises the low efficiency caused by using only ETPM to measure reliability in the star model. Finally, a prototype system is given, and the test results show that this method has built a trusted computing environment for embedded applications on existing hardware and software resources without additional hardware.

The existing trustworthiness attestation methods are not only difficult to be applied to the embedded system because they are mainly based on virtual machine technology, but have some problems such shat evidence is not obtained in time, protecting the privacy need trusted third party and trust measurement efficiency is low. In this paper, an embedded system dynamic trustworthiness attestation method based on dual-kernel (super kernel and normal kernel) operating system architecture is proposed. Super kernel is non-changeable, and it verifies the integrity of the critical data structures and kernel file in normal kernel. Super kernel can serve as a trusted third party which can dynamically verify whether the code segment changes in runtime. A system implementation is given in this paper, and the experimental data show that the behavior of the system can dynamically verify the behavior of program whether meets embedded trusted application demand or not.

Clustering is one of the main tasks used in pattern recognition and classification. Out of many methods that have been reported till date the most widely used methods are based on likelihood approach of mixture model. Among different mixture models, Expectation Maximization for Gaussian Mixture is most exploited and trusted algorithm for data clustering. However, it has some short comings such as initial parameters are to be given a-priori, convergence speed is slow and the results obtained are highly dependent upon the initial parameters. Many variations have been carried out in implementing EM algorithm but still there is ample scope for improvement. The proposed algorithm tries to overcome these shortcomings and provide more robust and efficient version of clustering algorithm. An improvement related to cluster partitioning is proposed in the existing algorithm resulting some advantages. The robustness and efficacy of the algorithm is demonstrated qualitatively as well as quantitatively with the help of some experiments.

Support vector machine (SVM) is a kind of machine learning method, but the selection of parameters has important effects on the generalization ability of SVMs. In this study, the relation between the error penalty parameter C, kernel parameter σ and the generalization ability of SVMs is discussed. Parameter C adjusts the similarity among within-class members, while parameter σ adjusts the similarity between classes. Moreover, C and σ balances each other mutually within a certain range, which forms a fan-shaped optional parameter distribution region. The optimal parameter area should be located near the center of the sector where both C and σ are small. According to this, a method is suggested to first search a suitable area with coarse grids, and then determine the optimal parameter within the area with a fine bilinear grid. Experimental results show that the new parameter selection method can not only avoid local optima, and thus excluding the cases in which C and σ are big and unstable, but also can be extremely fast in searching process. Compared with other parameter selection methods, the performance of SVMs cannot be influenced, or even better in some cases.

많은 양의 데이터로부터 유용성있는 정보의 추출, 진단 및 예후에 대한 결정, 질병 치료의 응용 등은 바이오인포머틱스(Bioinformatics)분야에서 매우 중요한 문제들이다. 본 논문에서는 암진단시스템에 적용하기위해 support vector machine을 위한 weighted kernel fuction과 빠른 수렴성과 좋은 분류성능을 갖는 학습방법을 제안하였다. 제안된 kernel function에서 기본적인 kernel fuction의 weights는 암진단 학습단계에서 결정되고 분류단계에서 파리미터로 사용된다. 대장암 데이터와 같은 임상 데이터에 대한 실험결과에서 제안된 방법은 기존의 다른 kernel fuction들 보다 더 우수하고 안정적인 분류성능을 보여주었다.

One of the most important problems in bioinformatics is how to extract the useful information from a huge amount of data, and make a decision in diagnosis, prognosis, and medical treatment applications. This paper proposes a weighted kernel function for support vector machine and its learning method with a fast convergence and a good classification performance. We defined the weighted kernel function as the weighted sum of a set of different types of basis kernel functions such as neural, radial, and polynomial kernels, which are trained by a learning method based on genetic algorithm. The weights of basis kernel functions in proposed kernel are determined in learning phase and used as the parameters in the decision model in classification phase. The experiments on several clinical datasets such as colon cancer indicate that our weighted kernel function results in higher and more stable classification performance than other kernel functions

The autoregressive process is applied in this paper to kernel regression in order to infer nonlinear models for predicting responses. We propose a kernel method for the autoregressive data which estimates the mean function by kernel machines. We also present the model selection method which employs the cross validation techniques for choosing the hyper-parameters which affect the performance of kernel regression. Artificial and real examples are provided to indicate the usefulness of the proposed method for the estimation of mean function in the presence of autocorrelation between data.

In this paper, the tenth-order linear boundary value problems are solved using reproducing kernel method. The algorithm developed approximates the solutions, and their higher-order derivatives, of differential equations and it avoids the complexity provided by other numerical approaches. First a new reproducing kernel space is constructed to solve this class of tenth-order linear boundary value problems; then the approximate solutions of such problems are given in the form of series using the present method. Three examples compared with those considered by Siddiqi, Twizell and Akram [S.S. Siddiqi, E.H. Twizell, Spline solutions of linear tenth order boundary value problems, Int. J. Comput. Math. 68 (1998) 345-362; S.S.Siddiqi, G.Akram, Solutions of tenth-order boundary value problems using eleventh degree spline, Applied Mathematics and Computation 185 (1)(2007) 115-127] show that the method developed in this paper is more efficient.

In this paper we propose a doubly penalized kernel method which estimates both the mean function and the variance function simultaneously by kernel machines for heteroscedastic autoregressive data. We also present the model selection method which employs the cross validation techniques for choosing the hyper-parameters which aect the performance of proposed method. Simulated examples are provided to indicate the usefulness of proposed method for the estimation of mean and variance functions.

We use the heat kernel method to prove newly the Paley-Wiener theorem for the distributions with compact support.

Background MicroRNAs (miRNAs) are a class of noncoding RNAs found in various organisms such as plants and mammals. However, most of the mRNAs regulated by miRNAs are unknown. Furthermore, miRNA targets in genomes cannot be identified by standard sequence comparison since their complementarity to the target sequence is imperfect in general. In this paper, we propose a kernel-based method for the efficient prediction of miRNA targets. To help in distinguishing the false positives from potentially valid targets, we elucidate the features common in experimentally confirmed targets. Results The performance of our prediction method was evaluated by five-fold cross-validation. Our method showed 0.64 and 0.98 in sensitivity and in specificity, respectively. Also, the proposed method reduced the number of false positives by half compared with TargetScan. We investigated the effect of feature sets on the classification of miRNA targets. Finally, we predicted miRNA targets for several miRNAs in the Caenorhabditis elegans (C. elegans) 3' untranslated region (3' UTR) database. Condusions The targets predicted by the suggested method will help in validating more miRNA targets and ultimately in revealing the role of small RNAs in the regulation of genomes. Our algorithm for miRNA target site detection will be able to be improved by additional experimental­knowledge. Also, the increase of the number of confirmed targets is expected to reveal general structural features that can be used to improve their detection.