Catálogo de publicaciones - libros

In this paper, we consider the problem of regrouping service sites into a smaller number of service sites called . Each service site is represented as a point in the plane and has service demand. We aim to group the sites so that each group has balanced service demand and the sum of distances between sites and their corresponding center is minimized. By using Voronoi diagrams, we obtain topological information among the sites and based on this, we define a mutation operator of a genetic algorithm. The experimental results show improvements in running time as well as cost optimization. We also provide a variety of empirical results by changing the relative importance of the two criteria, which involve service demand and distances, respectively.

In this paper we study a capacitated version of the classical max-k-cut problem. Given a graph we are to partition the vertices into equal-sized sets such that the number of edges across the sets is maximized. We present a deterministic approximation algorithm for this problem with performance ratio ( – 1) / . Our algorithm is based on local search, a technique that has been applied to a variety of combinatorial optimization problems.

In this paper we present the experimentation system with database for efficient job performing, using as an example an animal clinic. The structure of tasks and operations is based on the real procedures performed at an animal clinic. The assumed model implemented in the system is more sophisticated than a typical job-shop because of a multitude of various parameters, such as fatigue factor and the presence of uncertainty. The used heuristic priority algorithm enables us to analyze easily the impact of varied factors on the produced work-plan, and emphasize or switch off the impact of any considered factor. The system has been implemented in Matlab environment. In this work we present the opportunities of the proposed system on two examples. The first example of research is focused on choosing the optimal subset of performers for a given work. The second example concerns some work-rest policies, needed for evaluating the optimal work-rest model.

Radio Frequency Identification (RFID) has recently played an important role in ubiquitous sensing technology. While more advanced applications have been equipped with RFID devices, sensing multiple passive tags simultaneously becomes especially important. In this paper, using complementarily two-functioned estimation, we propose an identification method based on the stochastic process. The underlying mathematical principles and parameters estimation models have also been well discussed. Numerical examples are given to verify the proposed two-functioned estimation identification method within a given expected accuracy-level. Key Words: RFID (Radio Frequency Identification), Anti-collision Algorithm, Two-functioned Estimation.

We propose a proximal approach for solving a wide class of minimax location problems which in particular contains the round trip location problem. We show that a suitable reformulation of the problem allows to construct a Fenchel duality scheme the primal-dual optimality conditions of which can be solved by a proximal algorithm. This approach permits to solve problems for which distances are measured by mixed norms or gauges and to handle a large variety of convex constraints. Several numerical results are presented.

We consider the problem of determining the disassembly schedule (quantity and timing) of products in order to satisfy the demand of their parts or components over a finite planning horizon. This paper focuses on the capacitated version of the problem for the objective of minimizing the sum of setup, disassembly operation, and inventory holding costs. The problem is formulated as an integer program, and to solve the problem, a Lagrangean heuristic algorithm is developed after reformulating the integer program. To show the performance of the heuristic algorithm, computational experiments are done on randomly generated test problems, and the test results show that the algorithm suggested in this paper works well.

Biochemical reaction theory based DNA computation is of much better performance in solving a class of intractable computational problems such as NP-complete problems, it is important to study the DNA computation. A novel algorithm based on DNA computation is proposed, which solves a special category of 0-1 planning problem by using the surface-based fluorescence labeling technique. The analysis show that our new algorithm is of significant advantages such as simple encoding, low cost and short operating time, etc.

Clustering technique is essential for fast retrieval in large database. In this paper, new image clustering technique is proposed for content-based image retrieval. Fuzzy-ART mechanism maps high-dimensional input features into the output neuron. Joint HSV histogram and average entropy computed from gray-level co-occurrence matrices in the localized image region is employed as input feature elements. Original Fuzzy-ART suffers unnecessary increase of the number of output neurons when the noise input is presented. Our new Fuzzy-ART mechanism resolves the problem by differently updating the committed node and uncommitted node, and checking the vigilance test again. To show the validity of our algorithm, experiment results on image clustering performance and comparison with original Fuzzy-ART are presented in terms of recall rates.

It is well known that World Wide Web has become a huge information resource. The semi-structured data appears in a wide range of applications, such as digital libraries, on-line documentations, electronic commerce. After we have obtained enough data from WWW, we then use data mining method to mine schema knowledge from the data. Therefore, it is very important for us to utilize schema information effectively. This paper proposes a method of schema mining based on fuzzy decision tree to get useful schema information on the web. This algorithm includes three stages, represented using Datalog, incremental clustering, determining using fuzzy decision tree. Using this algorithm, we can discover schema knowledge implicit in the semi-structured data. This knowledge can make users understand the information structure on the web more deeply and thoroughly. At the same time, it can also provide a kind of effective schema for the querying of web information. In the future, we will further the work on extract association rules using machine learning method and study the clustering method in semi-structured data knowledge discovery.

A 3D Finite Difference Method (FDM) with spatially irregular grids is developed to simulate the seismic propagation in anisotropic media. Staggered irregular grid finite difference operator with second-order time and spatial accuracy are used to approximate the velocity-stress elastic wave equations. The parallel codes are implemented with Message Passing Interface (MPI) library and c language. The 3D model with complex earth structure geometry is split into more flexible subdomains by the proposed irregular method. The spurious diffractions from “staircase” interfaces can be easily eliminated without grid densification and costs less computing time. Parallel simulation scheme is described by pseudo codes. The spatial parallelism on PC cluster makes it a promising method for geo-science numerical computing. Parallel computation shows that the message passing between different CPUs are composed of the subdomain boundary information and need a considerable communication. The excellent parallelism speedup can be achieved through reasonable subdomain division and fast network connection.