Hierarchical multi-label classification using local neural networks

Hierarchical multi-label classification is a complex classification task where the classes involved in the problem are hierarchically structured and each example may simultaneously belong to more than one class in each hierarchical level. In this paper, we extend our previous works, where we investigated a new local-based classification method that incrementally trains a multi-layer perceptron for each level of the classification hierarchy. Predictions made by a neural network in a given level are used as inputs to the neural network responsible for the prediction in the next level. We compare the proposed method with one state-of-the-art decision-tree induction method and two decision-tree induction methods, using several hierarchical multi-label classification datasets. We perform a thorough experimental analysis, showing that our method obtains competitive results to a robust global method regarding both precision and recall evaluation measures.

Immbolization of uricase enzyme in Langmuir and Langmuir-Blodgett films of fatty acids: Possible use as a uric acid sensor

Preserving the enzyme structure in solid films is key for producing various bioelectronic devices, including biosensors, which has normally been performed with nanostructured films that allow for control of molecular architectures. In this paper, we investigate the adsorption of uricase onto Langmuir monolayers of stearic acid (SA), and their transfer to solid supports as Langmuir Blodgett (LB) films. Structuring of the enzyme in beta-sheets was preserved in the form of 1-layer LB film, which was corroborated with a higher catalytic activity than for other uricase-containing LB film architectures where the beta-sheets structuring was not preserved. The optimized architecture was also used to detect uric acid within a range covering typical concentrations in the human blood. The approach presented here not only allows for an optimized catalytic activity toward uric acid but also permits one to explain why some film architectures exhibit a superior performance. (C) 2011 Elsevier Inc. All rights reserved.

Solvent-Resistant Polymeric Sensors for Adulteration Detection in Liquid Fuels

Polymeric sensors with improved resistance to organic solvents were produced via the layer-by-layer thin film deposition followed by chemical cross-linking. According to UV-vis spectroscopy, the mass loss of polyaniline/poly(vinyl alcohol) and polyaniline/novolac-type resin based films deposited onto glass slides was less than 20% when they were submitted to successive immersions (up to 3,000 immersion cycles) into commercially available ethanol and gasoline fuel samples. Polyallylamine hydrochloride/nickel tetrasulfonated phthalocyanine films presented similar stability. The electrical responses assessed by impedance spectroscopy of films deposited onto Au-interdigitated microelectrodes were relatively unaffected after continuous or cyclic immersions into both fuels. After these studies, an array including these polymeric sensors was employed to detect adulteration in ethanol and gasoline samples. After principal component analysis, it was possible to conclude that the proposed sensor array is capable to discriminate with remarkable reproducibility ethanol samples containing different amounts of water or else gasoline samples containing different amounts of ethanol. In both examples, more than 90% of data variance was retained in the first principal component. For each type of sample, ethanol and gasoline, it was found a linear correlation between one of the principal components and the sample's composition. These findings allow one to conclude that these films present great potential for the development of reliable and low-cost sensors for fuel analysis in liquid phase.

Redes neurais artificiais na previsão da inflação: aplicação como ferramenta de apoio à análise de decisões financeiras em organizações de pequeno porte

As estimações das taxas de inflação são de fundamental importância para os gestores, pois as decisões de investimento estão intimamente ligadas a elas. Contudo, o comportamento inflacionário tende a ser não linear e até mesmo caótico, tornando difícil a sua correta estimação. Essa característica do fenômeno pode tornar imprecisos os modelos mais simples de previsão, acessíveis às pequenas organizações, uma vez que muitos deles necessitam de grandes manipulações de dados e/ou softwares especializados. O presente artigo tem por objetivo avaliar, por meio de análise formal estatística, a eficácia das redes neurais artificiais (RNA) na previsão da inflação, dentro da realidade de organizações de pequeno porte. As RNA são ferramentas adequadas para mensurar os fenômenos inflacionários, por se tratar de aproximações de funções polinomiais, capazes de lidar com fenômenos não lineares. Para esse processo, foram selecionados três modelos básicos de redes neurais artificiais Multi Layer Perceptron, passíveis de implementação a partir de planilhas eletrônicas de código aberto. Os três modelos foram testados a partir de um conjunto de variáveis independentes sugeridas por Bresser-Pereira e Nakano (1984), com defasagem de um, seis e doze meses. Para tal, foram utilizados testes de Wilcoxon, coeficiente de determinação R² e o percentual de erro médio dos modelos. O conjunto de dados foi dividido em dois, sendo um grupo usado para treinamento das redes neurais artificiais, enquanto outro grupo era utilizado para verificar a capacidade de predição dos modelos e sua capacidade de generalização. Com isso, o trabalho concluiu que determinados modelos de redes neurais artificiais têm uma razoável capacidade de predição da inflação no curto prazo e se constituem em uma alternativa razoável para esse tipo de mensuração.

Influence of Molecular Dynamics on the Dielectric Properties of Poly(9,9-di-n-octylfluorene-alt-benzothiadiazole) -Based Devices

This paper uses Nuclear Magnetic Resonance (NMR) and Differential Scanning Calorimetry (DSC) techniques to study the molecular relaxations and phase transitions in poly(9,9-di-n-octylfluorene-alt-benzothiadiazole) (F8BT), which has been extensively studied as the active thin film in organic devices. Besides the identification of the glass transition, beta relaxation and crystal-to-crystal phase transition, we correlate such phenomena with dielectric and transport mechanisms in diodes with F8BT as the active layer. The beta relaxation has been assigned to a transition at about 210 K measured by H-1 and C-13 solid state NMR, and can be attributed to local motions in the side chains. The glass transition has been detected by DSC and H-1 NMR. Dielectric spectroscopy (DS) carried out at low frequencies on diodes made from F8BT show two peaks which are coincident with the above transitions. This allowed us to correlate the electrical changes in the film with the onset of specific molecular motions. In addition, DS indicates a third peak related with a crystal-to-crystal phase transition. Finally, these transitions were correlated with changes in the carrier mobility recorded in thin films and published recently.

Spectroscopic, morphological and electrochromic characterization of layer-by-layer hybrid films of polyaniline and hexaniobate nanoscrolls

The combination of semiconducting oxides and polyaniline in the nanoscale range may result in hybrid materials having enhanced properties, such as electrochromism and charge capacity. This paper reports the spectroscopic, morphological and electrochromic characterization of hybrid films made up of hexaniobate one-dimensional (1D) nanoscrolls and polyaniline prepared by the layer-by-layer assembly technique (LbL). Secondary electron imaging and backscattered electron imaging techniques performed using a scanning electron microscope showed that polyaniline is adsorbed on the hexaniobate nanoscrolls, which confirms the combination of the components in the nanoscale domain. UV-VIS-NIR electronic spectra of the LbL hybrid films showed the absorption tail in the NIR region, assigned to delocalized polarons of the polyaniline. Resonance Raman spectra in the 1000-1700 cm(-1) range indicated that hybrid films present a higher relative intensity of polaron bands at 1337 and 1508 cm(-1) than pristine polyaniline in the emeraldine salt form. These results suggest that hexaniobate nanoscrolls induce a secondary doping of polyaniline. The cyclic voltammetry (CV) data for the hybrid film showed a specific capacity of 870 C cm(-3). According to CV results, the synergistic effect on charge storage properties of the hybrid material is attributed to the enhanced electroactivity of the hexaniobate component in the LbL film. Spectroelectrochemical experiments showed that the electrochromic efficiencies at 420 nm are ca. -41 and 24 cm(2) C-1 as the potential changes from 0.8 to -0.9 V and from -0.9 to -1.8 V, respectively, whereas at 800 nm the efficiencies are ca. -55 and 8 cm(2) C-1 for the same potential ranges. The electrochromic efficiencies and multi-colour character of the LbL film of hexaniobate nanoscrolls and polyaniline indicate that this novel hybrid material is an interesting modified electrode for electrochromic devices.

Dielectric properties of cobalt ferrite nanoparticles in ultrathin nanocomposite films

Multilayered nanocomposite films (thickness 50-90 nm) of cobalt ferrite nanoparticles (np-CoFe2O4, 18 nm) were deposited on top of interdigitated microelectrodes by the layer-by-layer technique in order to study their dielectric properties. For that purpose, two different types of nanocomposite films were prepared by assembling np-CoFe2O4 either with poly(3,4-ethylenedioxy thiophene):poly(styrene sulfonic acid) or with polyaniline and sulfonated lignin. Despite the different film architectures, the morphology of both was dominated by densely-packed layers of nanoparticles surrounded by polyelectrolytes. The dominant effect of np-CoFe2O4 was also observed after impedance spectroscopy measurements, which revealed that dielectric behavior of the nanocomposites was largely influenced by the charge transport across nanoparticle-polyelectrolyte interfaces. For example, nanocomposites containing np-CoFe2O4 exhibited a single low-frequency relaxation process, with time constants exceeding 15 ms. At 1 kHz, the dielectric constant and the dissipation factor (tan ᵟ) of these nanocomposites were 15 and 0.15, respectively. These values are substantially inferior to those reported for pressed pellets made exclusively of similar nanoparticles. Impedance data were further fitted with equivalent circuit models from which individual contributions of particle's bulk and interfaces to the charge transport within the nanocomposites could be evaluated. The present study evidences that such nanocomposites display a dielectric behavior dissimilar from that exhibited by their individual counterparts much likely due to enlarged nanoparticle- polyelectrolyte interfaces.