Uma Aplicação de Redes Neurais Auto-Organizaveis à Reconstrução Tridimensional de Superfícies

We propose a multi-resolution approach for surface reconstruction from clouds of unorganized points representing an object surface in 3D space. The proposed method uses a set of mesh operators and simple rules for selective mesh refinement, with a strategy based on Kohonen s self-organizing map. Basically, a self-adaptive scheme is used for iteratively moving vertices of an initial simple mesh in the direction of the set of points, ideally the object boundary. Successive refinement and motion of vertices are applied leading to a more detailed surface, in a multi-resolution, iterative scheme. Reconstruction was experimented with several point sets, induding different shapes and sizes. Results show generated meshes very dose to object final shapes. We include measures of performance and discuss robustness.

Identificação remota de sistemas operacionais utilizando análise de processos aleatórios e redes neurais artificiais

A new method to perform TCP/IP fingerprinting is proposed. TCP/IP fingerprinting is the process of identify a remote machine through a TCP/IP based computer network. This method has many applications related to network security. Both intrusion and defence procedures may use this process to achieve their objectives. There are many known methods that perform this process in favorable conditions. However, nowadays there are many adversities that reduce the identification performance. This work aims the creation of a new OS fingerprinting tool that bypass these actual problems. The proposed method is based on the use of attractors reconstruction and neural networks to characterize and classify pseudo-random numbers generators

Uma proposição para o cálculo de mapas de disparidade de imagens estéreo usando um interpolador neural baseado em funções de base radial

This study aims to seek a more viable alternative for the calculation of differences in images of stereo vision, using a factor that reduces heel the amount of points that are considered on the captured image, and a network neural-based radial basis functions to interpolate the results. The objective to be achieved is to produce an approximate picture of disparities using algorithms with low computational cost, unlike the classical algorithms

Utilização de redes neurais artificiais para detecção e diagnóstico de falhas

In a real process, all used resources, whether physical or developed in software, are subject to interruptions or operational commitments. However, in situations in which operate critical systems, any kind of problem may bring big consequences. Knowing this, this paper aims to develop a system capable to detect the presence and indicate the types of failures that may occur in a process. For implementing and testing the proposed methodology, a coupled tank system was used as a study model case. The system should be developed to generate a set of signals that notify the process operator and that may be post-processed, enabling changes in control strategy or control parameters. Due to the damage risks involved with sensors, actuators and amplifiers of the real plant, the data set of the faults will be computationally generated and the results collected from numerical simulations of the process model. The system will be composed by structures with Artificial Neural Networks, trained in offline mode using Matlab®

Contribuições a técnicas de agrupamento e visualização de dados multivariados utilizando mapas auto-organizáveis

Self-organizing maps (SOM) are artificial neural networks widely used in the data mining field, mainly because they constitute a dimensionality reduction technique given the fixed grid of neurons associated with the network. In order to properly the partition and visualize the SOM network, the various methods available in the literature must be applied in a post-processing stage, that consists of inferring, through its neurons, relevant characteristics of the data set. In general, such processing applied to the network neurons, instead of the entire database, reduces the computational costs due to vector quantization. This work proposes a post-processing of the SOM neurons in the input and output spaces, combining visualization techniques with algorithms based on gravitational forces and the search for the shortest path with the greatest reward. Such methods take into account the connection strength between neighbouring neurons and characteristics of pattern density and distances among neurons, both associated with the position that the neurons occupy in the data space after training the network. Thus, the goal consists of defining more clearly the arrangement of the clusters present in the data. Experiments were carried out so as to evaluate the proposed methods using various artificially generated data sets, as well as real world data sets. The results obtained were compared with those from a number of well-known methods existent in the literature

Excitações elementares em super-redes quasiperiódicas com simetria espelho

In this thesis, we investigated the magnonic and photonic structures that exhibit the so-called deterministic disorder. Speci cally, we studied the effects of the quasiperiodicity, associated with an internal structural symmetry, called mirror symmetry, on the spectra of photonics and magnonics multilayer. The quasiperiodicity is introduced when stacked layers following the so-called substitutional sequences. The three sequences used here were the Fibonacci sequence, Thue-Morse and double-period, all with mirror symmetry. Aiming to study the propagation of light waves in multilayer photonic, and spin waves propagation in multilayer magnonic, we use a theoretical model based on transfer matrix treatment. For the propagation of light waves, we present numerical results that show that the quasiperiodicity associated with a mirror symmetry greatly increases the intensity of transmission and the transmission spectra exhibit a pro le self-similar. The return map plotted for this system show that the presence of internal symmetry does not alter the pattern of Fibonacci maps when compared with the case without symmetry. But when comparing the maps of Thue-Morse and double-time sequences with their case without the symmetry mirror, is evident the change in the pro le of the maps. For magnetic multilayers, we work with two di erent systems, multilayer composed of a metamagnetic material and a non-magnetic material, and multilayers composed of two cubic Heisenberg ferromagnets. In the rst case, our calculations are carried out in the magnetostatic regime and calculate the dispersion relation of spin waves for the metamgnetic material considered FeBr2. We show the e ect of mirror symmetry in the spectra of spin waves, and made the analysis of the location of bulk bands and the scaling laws between the full width of the bands allowed and the number of layers of unit cell. Finally, we calculate the transmission spectra of spin waves in quasiperiodic multilayers consisting of Heisenberg ferromagnets. The transmission spectra exhibit self-similar patterns, with regions of scaling well-de ned in frequency and the return maps indicates only dependence of the particular sequence used in the construction of the multilayer