Catálogo de publicaciones - libros

Interflation of the function (x,……,x) of the $n$ variables with help of the its traces (and traces of its derivatives of order ≤ N) on the surfaces the dimension is recovery (possible, exactly) f ., If this is interpolation on points (for n ≥ 1). If (for ) it is interlineation (blending function interpolation) on lines. In this paper the review of last achievements and some applications interflation, interlineation functions and blending approximation functions for construction the economical algorithms in the approximation theory is given.

Independent component analysis (ICA) has attracted much interest in dynamic contrast-enhanced (DCE) neuroimaging, as it allows for blind separation of the brain haemodynamic patterns (components). However, the exact number of components is always unknown in practice. In this work, numerical simulation study was carried out to compare the performance of ICA by using the principle component analysis (PCA) to reduce the number of components. Oscillatory indices method and kurtosis method are also discussed for automatically selecting the components of interest.

We transform the geometric constraint solving into the numerical optimization solving. A new hybrid algorithm is proposed which combines the merits of global search of the genetic algorithm (GA) and the good property of local search of the conjugate gradient approach. This algorithm uses GA to search the area where the best solution may exist in the whole space, and then performs fine searching. When the algorithm approaches to the best solution and the search speed is too slow, we can change to the effective local search strategy—the conjugate gradient algorithm in order to enhance the ability of the GA on fine searching. It makes the algorithm get rid off the prematurity convergence situation. We apply this algorithm into the geometric constraint solving. The experiment shows that the hybrid algorithm has the effective convergence property and it can find the global best solution.

In this paper, we propose a new perceptual digital watermarking scheme based on ICA-SCS denoising and the orthogonal vectors. In the watermarking detection process, the independent component analysis (ICA)-based on sparse code shrinkage (SCS) technique is employed to denoise and making using of the orthogonal vectors’ property. By hypothetical testing, the watermark can be extracted exactly. The experimental results show that our method successfully survives image processing operation, image cropping, noise adding and the JPEG lossy compression. Especially, the scheme is robust towards noise adding, image sharpening and image enhancement.

Web pages are defined by the HTML. Web pages can be linked by hyper links, so as to construct a hyper media system. For web mining, there must be a data model to describe the rules of interest relation among web pages. In this paper, we simplify the WWW data model that can be used for web mining.

The procedure of process models executing in workflow engine is described using operational semantic. The formalized description of process instances and activity instances lead directly to the realization of workflow engine. Therefore, it provides a unified processing procedure for the realization of the workflow engine software, which makes it convenient for the cooperation among heterogeneous workflow engines.

We have developed an interactive Web application implemented by Java Server Pages technology for executing the fuzzy c-means clustering method on the input data given by users from client side. On the server side the Tomcat application server dispatches the requests, produces the answers and presents them on the client side. The server machine is not a robust computer (Intel Pentium II category). All of the instaled softwares are under free licence policy.

Although many methods for feature selection and text classification have been applied to English web documents, relatively few studies have been done on Chinese web documents. This paper introduces a term weighting method based on inverse document frequency, html tags and length of Chinese phrase, provides an algorithm for web text classification based on improving on lattice machine approach. The experiments show this method is effective in feature reduction and text classification.

Resource-constrained scheduling problem is one kind of typical real-life discrete optimization problems, which is one of the strongest application areas of constraint programming. In the constraint programming toolkit ‘Mingyue’, which embed constraints in the object-oriented language C++, we design a new logic-based method for handling the constraints in the resource-constrained scheduling problem. In this paper, we propose a way of describing those constraints with the discrete-variable logic formula. Based on this model, a resolution algorithm is designed for filtering the discrete variables’ domain. Comparisons with other constraint handling approaches and related literature clearly show that our approach can describe the constraints in the high level and solve the resource-constrained scheduling problem in the logic framework.

A topic driven crawler chooses the best URLs to pursue during web crawling. It is difficult to evaluate what URLs downloaded are the best. This paper presents some important metrics and an evaluation function for ranking URLs about pages relevance. We also discuss an approach to evaluate the function based on GA. The best combination of the metrics’ weights can be discovered by GA evolving process. The experiment shows that the performance is exciting, especially about a popular topic.