Catálogo de publicaciones - libros

Breast cancer is one of the major tumor related cause of death in women. Various artificial intelligence techniques have been used to improve the diagnoses procedures and to aid the physician’s efforts. In this paper we summarize our preliminary study to detect breast cancer using a Flexible Neural Tree (FNT), Neural Network (NN), Wavelet Neural Network (WNN) and their ensemble combination. For the FNT model, a tree-structure based evolutionary algorithm and the Particle Swarm Optimization (PSO) are used to find an optimal FNT. For the NN and WNN, the PSO is employed to optimize the free parameters. The performance of each approach is evaluated using the breast cancer data set. Simulation results show that the obtained FNT model has a fewer number of variables with reduced number of input features and without significant reduction in the detection accuracy. The overall accuracy could be improved by using an ensemble approach by a voting method.

The development of techniques for a navigation of multiple mobile robots is abroad topic, covering a large spectrum of different technologies and applications. Neural networks and fuzzy logic control techniques can improve real-time control performance for a mobile robot due to their high robustness and error-tolerance ability. This paper proposes a neuro-fuzzy (NF) controller, which integrates the transparency of the fuzzy logic with the learning capability of neural networks is developed for multiple mobile robots navigation in an unknown environment. The neuro-fuzzy controller developed in this research consists of a neural network pre-processor followed by a fuzzy logic controller. The former is structured using multi-layer perceptron (MLP) or local model network (LMN). Practical results reflect the soundness of the proposed scheme.

This work proposes an integrated rough genetic approach with modified definitions of rough set approximation for knowledge discovery from large and complex data bases. Rough set theory has been used for attribute selection while genetic algorithm has been used for finding out the right set of compact rules that covers most of the objects of the data base. The simulation results with moderately large data base have been found to be promising.

The purpose of this study is to clarify the interaction between the formation mechanism of the distribution pattern of ants supported by pheromone signals and the regulation mechanism of task-allocation in collective behavior of an ant colony. In this paper I design two types of ant colony model focusing on pheromone signals and the sensitivity of ants, and analyze simulated behaviors of the models. First, I design three foraging models (trail, attraction and desensitization models) composed by ants with different sensitivities to recruit pheromone. Among them, the desensitization model composed for ants that change their sensitivities in response to environmental information around them always shows the best foraging efficiency, stable recruitment pattern and balanced allocation among subtasks. In the second, I design a task-allocation model composed for ants carrying out foraging and mound-piling tasks using independent signals for each task. The results indicate that weak interference occurs between both tasks.

A generalized version of Gutjahr’s Graph-based Ant System (GBAS) framework for solving static combinatorial optimization problems is examined in the present paper. A new transition rule which intends to balance between the exploration and the exploitation in the search progress of Ant-based algorithms, is added into Gutjahr’s GBAS model. As shown in this paper, our generalized model still holds all convergent properties of the GBAS model and may show a promising improvement in solution quality to Ant-based algorithms in literature.

This paper presents an unsupervised data clustering method based on flowers pollination by artificial bees we named it FPAB. FPAB does not require any parameter settings and any initial information such as the number of classes and the number of partitions on input data. Initially, in FPAB, bees move the pollens and pollinate them. Each pollen will grow in proportion to its garden flowers. Better growing will occur in better conditions. After some iteration natural selection reduces the pollens and flowers to form gardens of same type of flowers. The prototypes of each gardens are taken as the initial cluster centers for Fuzzy C Means algorithm which is used to reduce obvious misclassification errors. In the next stage the prototypes of gardens are assumed as a single flower and FPAB is applied to them again. Results from three small data sets show that the partitions produced by FPAB are competitive with those obtained from FCM or AntClass.

In this study, ε constrained particle swarm optimizer εPSO, which is the combination of the ε constrained method and particle swarm optimization, is proposed to solve constrained optimization problems. The ε constrained methods can convert algorithms for unconstrained problems to algorithms for constrained problems using the ε level comparison, which compares the search points based on the constraint violation of them. In the ε PSO, the agents who satisfy the constraints move to optimize the objective function and the agents who don’t satisfy the constraints move to satisfy the constraints. Also, the way of controlling ε-level is given to solve problems with equality constraints. The effectiveness of the ε PSO is shown by comparing the ε PSO with GENOCOP5.0 on some nonlinear constrained problems with equality constraints.

Many applications arising from areas of economics, finance and engineering are cast into the sum-of-ratios problem. Usually, the problems are on such a large scale that the existing algorithms are naive yet to obtain an optimal solution of the problems. In this study we develop a heuristic algorithm to obtain such a better solution of the sum-of-ratios problem by means of Ant Colony System. The proposed algorithm can be used for designing a globally optimal algorithm with the help of some certain strategy of global search as well. We report numerical experiments of the heuristic algorithm, which indicates that the best function value obtained from our heuristic algorithm is empirically near to the optimal value with a high probability.

Warehouse management system (WMS) today is viewed as a basis to reinforcing company logistics. Order picking is one of the routine operations in warehouses. Before picking a large amount of orders, effectively grouping orders into batches can speed up product movement within the warehouse. Several batching heuristics have been proposed in the literature for minimizing travel distance or travel time. This paper presents an order batching approach in a distribution center with a parallel-aisle layout. A heuristic order batching approach based on data mining is developed in this paper.

Non-standard database applications such as CAD/CAM or geographic information processing are becoming increasingly important. Such application systems must be equipped with the capability of effective accessibility to spatial data. The spatial domain consists of many spatial objects that are made up of points, lines, regions, and even high dimensional data. In order to effectively manipulate the spatial data, the tree structure is applied. In this paper, we consider such problems as spatial data retrieval, dynamic manipulation and storage utilization by indexing the large spatial data. A new tree structure, Five- Area Tree (denotes to FA-Tree), is proposed to organize the spatial data. Also, our experimental results show that the FA-Tree has better storage utilization than the Nine-Area Tree (also known as the NA-Tree).