952 resultados para All-optical networks
Resumo:
Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
Resumo:
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Resumo:
Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
Resumo:
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Resumo:
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)
Resumo:
Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
Resumo:
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Resumo:
This article deals with classification problems involving unequal probabilities in each class and discusses metrics to systems that use multilayer perceptrons neural networks (MLP) for the task of classifying new patterns. In addition we propose three new pruning methods that were compared to other seven existing methods in the literature for MLP networks. All pruning algorithms presented in this paper have been modified by the authors to do pruning of neurons, in order to produce fully connected MLP networks but being small in its intermediary layer. Experiments were carried out involving the E. coli unbalanced classification problem and ten pruning methods. The proposed methods had obtained good results, actually, better results than another pruning methods previously defined at the MLP neural network area. (C) 2014 Elsevier Ltd. All rights reserved.
Resumo:
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)
Resumo:
We report, for the first time to our knowledge, experimental results on pedestal waveguides produced with Yb3+/Er3+ codoped Bi2O3-WO3-TeO2 thin films deposited by RF Sputtering for photonic applications. Thin films were deposited using Ar/O-2 plasma at 5 mTorr pressure and RF power of 40 W on substrates of silicon wafers. The definition of the pedestal waveguide structure was made using conventional optical lithography followed by plasma etching. Propagation losses around 2.0 dB/cm and 2.5 dB/cm were obtained at 633 and 1050 nm, respectively, for waveguides in the 20-100 mu m width range. Single-mode propagation was measured for waveguides width up to 10 mu m and 12 mu m, at 633 nm and 1050 nm, respectively; for larger waveguides widths multi-mode propagation was obtained. Internal gain of 5.6 dB at 1530 nm, under 980 nm excitation, was measured for 1.5 cm waveguide length (similar to 3.7 dB/cm). The present results show the possibility of using Yb3+/Er3+ codoped Bi2O3-WO3-TeO2 pedestal waveguide for optical amplifiers. (C) 2014 Elsevier B.V. All rights reserved.
Resumo:
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)
Resumo:
PURPOSE: To report a case of idiopathic macular hole, with vitreoretinal traction confirmed by optical coherence tomography that was successfully treated by a single intravitreous perfluoropropane (C3F8) gas bubble injection.METHODS: Case report. A 65-year-old patient with idiopathic macular hole (stage 2, one eye) received an intravitreous gas injection and was prospectively followed with optical coherence tomography.RESULTS: A complete posterior vitreous detachment was achieved within 6 weeks after gas injection. Visual acuity improved from 20/80 to 20/25 by 10 months of followup. Optical coherence tomography disclosed vitreoretinal traction release and macular hole closure. No complications were related to the procedure.CONCLUSION: This simple procedure can assist a complete posterior vitreous detachment with relief of the hyaloid-foveolar traction, facilitating macular hole closure. (C) 2001 by Elsevier Science Inc. All rights reserved.
Resumo:
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Resumo:
Background: Next-generation sequencing (NGS) allows for sampling numerous viral variants from infected patients. This provides a novel opportunity to represent and study the mutational landscape of Hepatitis C Virus (HCV) within a single host.Results: Intra-host variants of the HCV E1/E2 region were extensively sampled from 58 chronically infected patients. After NGS error correction, the average number of reads and variants obtained from each sample were 3202 and 464, respectively. The distance between each pair of variants was calculated and networks were created for each patient, where each node is a variant and two nodes are connected by a link if the nucleotide distance between them is 1. The work focused on large components having > 5% of all reads, which in average account for 93.7% of all reads found in a patient. The distance between any two variants calculated over the component correlated strongly with nucleotide distances (r = 0.9499; p = 0.0001), a better correlation than the one obtained with Neighbour-Joining trees (r = 0.7624; p = 0.0001). In each patient, components were well separated, with the average distance between (6.53%) being 10 times greater than within each component (0.68%). The ratio of nonsynonymous to synonymous changes was calculated and some patients (6.9%) showed a mixture of networks under strong negative and positive selection. All components were robust to in silico stochastic sampling; even after randomly removing 85% of all reads, the largest connected component in the new subsample still involved 82.4% of remaining nodes. In vitro sampling showed that 93.02% of components present in the original sample were also found in experimental replicas, with 81.6% of reads found in both. When syringe-sharing transmission events were simulated, 91.2% of all simulated transmission events seeded all components present in the source.Conclusions: Most intra-host variants are organized into distinct single-mutation components that are: well separated from each other, represent genetic distances between viral variants, robust to sampling, reproducible and likely seeded during transmission events. Facilitated by NGS, large components offer a novel evolutionary framework for genetic analysis of intra-host viral populations and understanding transmission, immune escape and drug resistance.
Resumo:
Complex non-linear interactions between banks and assets we model by two time-dependent Erdos-Renyi network models where each node, representing a bank, can invest either to a single asset (model I) or multiple assets (model II). We use a dynamical network approach to evaluate the collective financial failure -systemic risk- quantified by the fraction of active nodes. The systemic risk can be calculated over any future time period, divided into sub-periods, where within each sub-period banks may contiguously fail due to links to either i) assets or ii) other banks, controlled by two parameters, probability of internal failure p and threshold T-h ("solvency" parameter). The systemic risk decreases with the average network degree faster when all assets are equally distributed across banks than if assets are randomly distributed. The more inactive banks each bank can sustain (smaller T-h), the smaller the systemic risk -for some Th values in I we report a discontinuity in systemic risk. When contiguous spreading becomes stochastic ii) controlled by probability p(2) -a condition for the bank to be solvent (active) is stochasticthe- systemic risk decreases with decreasing p(2). We analyse the asset allocation for the U.S. banks. Copyright (C) EPLA, 2014