Adaptive covariate acquisition for minimizing total cost of classification

Machine Learning - In some applications, acquiring covariates comes at a cost which is not negligible. For example in the medical domain, in order to classify whether a patient has diabetes or not,...     

Secondary Metabolic Profile as a Tool for Distinction and Characterization of Cultivars of Black Pepper (Piper nigrum L.) Cultivated in Pará State,Brazil

This study evaluated the chemical compositions of the leaves and fruits of eight black pepper cultivars cultivated in Pará State (Amazon, Brazil). Hydrodistillation and gas chromatography–mass spectrometry were employed to extract and analyze the volatile compounds, respectively. Sesquiterpene hydrocarbons were predominant (58.5–90.9%) in the cultivars “Cingapura”, “Equador”, “Guajarina”, “Iaçará”, and “Kottanadan”, and “Bragantina”, “Clonada”, and “Uthirankota” displayed oxygenated sesquiterpenoids (50.6–75.0%). The multivariate statistical analysis applied using volatile composition grouped the samples into four groups: γ-Elemene, curzerene, and δ-elemene (“Equador”/“Guajarina”, I); δ-elemene (“Iaçará”/“Kottanadan”/“Cingapura”, II); elemol (“Clonada”/“Uthirankota”, III) and α-muurolol, bicyclogermacrene, and cubebol (“Bragantina”, IV). The major compounds in all fruit samples were monoterpene hydrocarbons such as α-pinene, β-pinene, and limonene. Among the cultivar leaves, phenolics content (44.75–140.53 mg GAE·g⁻¹ FW), the enzymatic activity of phenylalanine-ammonia lyase (20.19–57.22 µU·mL⁻¹), and carotenoids (0.21–2.31 µg·mL⁻¹) displayed significant variations. Due to black pepper’s susceptibility to Fusarium infection, a molecular docking analysis was carried out on Fusarium protein targets using each cultivar’s volatile components. F. oxysporum endoglucanase was identified as the preferential protein target of the compounds. These results can be used to identify chemical markers related to the susceptibility degree of black pepper cultivars to plant diseases prevalent in Pará State.     

Nitrogen dynamics during till and no-till pasture restoration sequences in Rondônia,Brazil

The clearing of tropical rain forest in the Amazon basin has created large areas of cattle pasture that are now declining in productivity. Practices adopted by ranchers to restore productivity to degraded pastures have the potential to alter soil N availability and gaseous N losses from soils. We examined how soil inorganic N pools, net N mineralization and net nitrification rates, nitrification potential and NO and N₂O emissions from soils of a degraded pasture responded to the following restoration treatments: (1) soil tillage followed by replanting of grass and fertilization, (2) no-till application of non-selective herbicide, planting of rice, harvest followed by no-till replanting of grass and fertilization, and (3) the same no-till sequence with soybeans instead of rice. Tillage increased soil NH₄⁺ and NO₃^? pools but NH₄⁺ and NO₃^? pools remained relatively constant in the control and no-till treatments. Cumulative rates of net N mineralization and net nitrification during the first 6 months after treatment varied widely but were hightest in the tilled treatment. Emissions of NO and N₂O fluxes increased with tillage and with N fertilization. There were no clear relationships among rates of N fertilizer application, net N mineralization, net nitrification, NO, N₂O and total N oxide emissions. Our results indicate that pasture restoration sequences involving tilling and fertilizing will increase emissions of N oxides, but the magnitude of the increase is likely to differ based on timing of fertilizer application relative to the presence of plants and the magnitude of plant N demand. Emissions of N oxides appear to be decreased by the use of restoration sequences that minimize reductions in pasture grass cover.     

Influence of the composition of the binder and the carbonation on the zeta potential values of hardened cementitious materials

Electroosmotic experiments have been carried out on eight different mixes of hardened cementitious matrixes including concrete, mortar and pastes, different types of binders, carbonated and non-carbonated specimens and different test conditions. From these trials, the zeta potential values have been determined and the optimum experimental parameters and devices have been established. In addition, the influence of the composition of the binder on the zeta potential has been quantified indicating that higher amounts of Al₂O₃ in the binder lead to higher positive values of zeta potential, while the SiO₂ acts just on the opposite side. This has been explained in terms of the different i.e.p. (isoelectric point) for both oxides. Concerning the influence of carbonation, it increases the absolute value of the zeta potential toward more negative values; this behaviour has been attributed to the influence of the specifically adsorbed Ca²⁺ in the negative sites of the surface in the non-carbonated samples.     

Land Use Effect on the CN Model Parameters in a Tropical Dry Environment

Dry tropical forests account for over 1,000,000 km², and there is still lack of knowledge on their hydrologic processes. The curve number (CN) hydrologic model developed by the Natural Resources Conservation Service (NRCS) is widely applied for runoff determination in various parts of the world, but not so in tropical semiarid regions. This study analyzes the impact of land use changes on the CN model in a tropical semiarid environment, in two catchments of native dry tropical forest and thinned dry tropical forest land use from 2009 to 2012. The CN model was calibrated and validated for the NRCS recommended initial abstraction ratio λ = 0.2, and for λ evaluated from rainfall and runoff data. A reliability analysis was performed using Monte Carlo simulation. Model goodness-of-fit was assessed with statistical criteria. A total of 42 and 40 rainfall-runoff events were analyzed for the native and thinned dry tropical forest, respectively. Characteristic λ values of 0.15 and 0.11 were determined for the two respective catchments. Although CN values were similar for both land uses, CN_λ=0.20 = 80 and CN_{median λ} = 77, the thinned catchment showed a higher CN model parameters variability. The CN model was more sensitive to variations of CN values than to those of λ. This study showed that no matter the vegetation management in a dry tropical forest environment, modeled runoff is not affected by λ, but rather affected by CN, which represents soil, landuse and management.     

Effect of the reaction conditions on the properties of poly(4,4′-diphenylether-1,3,4-oxadiazole)s

Summary Poly(4,4-diphenylether-1,3,4-oxadiazole)s have been prepared by polycondensation of 4,4-diphenylether dicarboxylic acid and hydrazine sulphate, HS, in poly(phosphoric acid), PPA, under different reaction condictions. The products were characterized by viscometry, gel permeation chromatography, ¹H-NMR spectrometry and thermogravimetric analysis. Dense membranes have been prepared and submitted to gas transport and wide angle X-ray diffraction, WAXD, experiments. The differences found in permeability and selectivity parameters, as well as in density and WAXD results, are discussed in terms of their solution viscosities and hydrazide repeat unit contents.     

Inductive Model Generation for Text Classification Using a Bipartite Heterogeneous Network

Algorithms for numeric data classification have been applied for text classification. Usually the vector space model is used to represent text collections. The characteristics of this representation such as sparsity and high dimensionality sometimes impair the quality of general-purpose classifiers. Networks can be used to represent text collections, avoiding the high sparsity and allowing to model relationships among different objects that compose a text collection. Such network- based representations can improve the quality of the classification results. One of the simplest ways to represent textual collections by a network is through a bipartite heterogeneous network, which is composed of objects that represent the documents connected to objects that represent the terms. Heterogeneous bipartite networks do not require computation of similarities or relations among the objects and can be used to model any type of text collection. Due to the advantages of representing text collections through bipartite heterogeneous networks, in this article we present a text classifier which builds a classification model using the structure of a bipartite heterogeneous network. Such an algorithm, referred to as IMBHN （Inductive Model Based on Bipartite Heterogeneous Network）, induces a classification model assigning weights to objects that represent the terms for each class of the text collection. An empirical evaluation using a large amount of text collections from different domains shows that the proposed IMBHN algorithm produces significantly better results than k-NN, C4.5, SVM, and Naive Bayes algorithms.     

Electrocatalytic oxidation of ethanol on Pt–Mo bimetallic electrodes in acid medium

Pt–Mo alloy electrocatalysts were prepared by an arc-melting furnace process to investigate the origin of their enhanced activity toward ethanol oxidation. Two Mo contents were chosen in zones of the binary phase diagram where they are supposed to form either a pure alloy mixture or a solid solution. Pt–Mo alloy catalysts were more active than Pt-alone. Gradual Mo dissolution at the electrode surface was observed after voltammetric and chronoamperometric measurements. The dissolved Mo contributed to the catalytic effect of the electrode as underpotentially deposited (upd) adatoms. This dissolution probably also leads to an increase in the electrode surface roughness. Low molybdenum content in the electrode material enhances the activity toward ethanol oxidation when compared to Pt-alone. Ethanol oxidation was also investigated by in situ infrared reflectance spectroscopy in order to determine the presence of adsorbed intermediates like CO species. Acetaldehyde, acetic acid and CO₂ were also found by spectroscopic experiments.     

Dielectric behavior of PVDF/POMA blends that have a low doped POMA content

The real (ε′) and imaginary (ε″) components of the complex permittivity of blends of PVDF [poly(vinylidene fluoride)] with POMA [poly(o‐methoxyaniline)] doped with toluenosulfonic acid (TSA) containing 1, 2.5, and 5 wt % POMA–TSA were determined in the frequency interval between 10² and 3 × 10⁶ Hz and in the temperature range from ?120 up to 120°C. It was observed that the values of ε′ and ε″ had a greater increase with the POMA–TSA content and with a temperature in the region of frequencies below 10 kHz. This effect decreased with frequency and it was attributed to interfacial polarization. This polarization was caused by the blend heterogeneity, formed by conductive POMA–TSA agglomerates dispersed in an insulating matrix of PVDF. The equation of Maxwell–Garnett, modified by Cohen, was used to evaluate the permittivity and conductivity behavior of POMA–TSA in the blends. A strong decrease was observed in POMA–TSA conductivity in the blend, which was bigger the lower the POMA–TSA content in the blend. This decrease could have been caused either by the POMA dedoping during the blend preparation process or by its dispersion into the insulating matrix. © 2002 Wiley Periodicals, Inc. J Appl Polym Sci 87: 752–758, 2003