Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-01-03 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the problems of log detection accuracy and long delay of DDoS(Distributed Denial-of-Service) attacks in the software definition networks, This paper proposed a real time DDoS security system of software definition networks based on kernel functions. Firstly, it abstracted the packet header fields of software definition networks periodly, and formed the abstracted information as matrices; then, it adopted the Mahalanobis distance to analyze the significant change of continuous feature vectors, and it designed two kernel functions to evaluate the behavior flows of attacks; lastly, the attackers are identified by the spectral clustering technique and the covariance statistical information. Experimental results based on the real software definition networks show that the proposed security system realizes a good detection accuracy, and performs a reasonable processing time.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-12-13 Cooperative journals: 《计算机应用研究》
Abstract： In order to solve the problem that the low utilization rate of the neighborhood information and spatial information led to vulnerability to noise of fuzzy clustering algorithm, This paper proposed a fuzzy C-means algorithm combining kernel function and Mahalanobis distance(FCMKM) . Firstly, it non-linearly mapped the image pixels from low-dimensional space to high-dimensional space through the kernel function. Then, it replaced the original Euclidean distance with Mahalanobis distance as a high-dimensional spatial distance measurement. Finally, it used the improved algorithm to segment the image. The paper selected five evaluation indexes Bezdek partition coefficient, Xie_Beni coefficient, reconstruction error rate, running time and iteration number as evaluation criteria of comparative experiments to verify the performance of the FCMKM algorithm. Experimental results show that compared with traditional FCM algorithm, kernel function based FCM algorithm and markov distance based FCM algorithm, FCMKM algorithm can effectively improve the anti-noise performance of fuzzy clustering algorithm.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-12-13 Cooperative journals: 《计算机应用研究》
Abstract： Due to that most high-dimensional data not only has the similarities, but also nonlinear relationships. This paper proposed a nonlinear feature selection algorithm based on local structure learning. Firstly, the algorithm mapped the data to high-dimensional space through kernel functions, and expresses the nonlinear relationship between data features in high-dimensional space. Then, it exploited the similarity between the features in the low-dimensional space through local structural learning. At the same time, it eliminated the interference of noise by the low-rank constraint. Finally, it selected features by sparse regularization factors. It find the non-linear relationships between data features by the kernel function, and find the similarities between the data attributes as the local structure learning. The algorithm is a nonlinear feature selection algorithm embedded with local structure learning. Experimental results show that the algorithm has better results than other comparison algorithms.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-20 Cooperative journals: 《计算机应用研究》
Abstract： A fast face verification method based on Gaussian process in small sample space was proposed to solve the problem of large training samples, complex computation and the slow recognition. Firstly, it used the conjugate gradient descent to detect the face feature points, and then used the adaptive multi-scale local binary model to extract the features at the feature points reduced the feature dimensions. Finally, the spectral kernel function is used as the kernel function of the Gaussian process to classify the input face features. In this paper, training and testing are carried out using LFW, FERET and Multi-PIE face database. The experimental results showed that the local binary model can effectively reduce the feature dimension. The combination of the Gaussian process model and the spectral hybrid core can greatly reduce the training samples and improve the training speed and test speed.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： Aiming at the problem that the traditional clustering algorithm can only deal with single attribute data and can’t handle the clustering problem of mixed type data very well. Most of the clustering algorithms for mixed type data currently have the problem of initializing sensitive and can’t handle the data of arbitrary shape. This paper proposed an entropy-based spectral clustering algorithm for mixed type data to deal with mixed type data. First, it proposed a new similarity measure. It used the numerical data in the spectral clustering algorithm constitutes a Gaussian kernel function of the matrix, and used the classification data constitutes an entropy-based the influence factor of the matrix. A new similarity matrix combines these two matrices. Instead of the traditional similarity matrix, it proposed the new similarity matrix avoid feature transformation and parameter adjustment between the numerical data and the classification data. Then, it applied the new similarity matrix to the spectral clustering algorithm so as to deal with the data of arbitrary shape, and finally got the clustering result. Experiments on UCI data sets show that this algorithm can effectively deal with the clustering problem of mixed attribute data, with high stability and good robustness.
Hits
YES
Hits
Downloads
Comment