A review on the combination of binary classifiers in multiclass problems

Several real problems involve the classification of data into categories or classes. Given a data set containing data whose classes are known, Machine Learning algorithms can be employed for the induction of a classifier able to predict the class of new data from the same domain, performing the desired discrimination. Some learning techniques are originally conceived for the solution of problems with only two classes, also named binary classification problems. However, many problems require the discrimination of examples into more than two categories or classes. This paper presents a survey on the main strategies for the generalization of binary classifiers to problems with more than two classes, known as multiclass classification problems. The focus is on strategies that decompose the original multiclass problem into multiple binary subtasks, whose outputs are combined to obtain the final prediction.     

Relating instance hardness to classification performance in a dataset: a visual approach

Machine Learning - Machine Learning studies often involve a series of computational experiments in which the predictive performance of multiple models are compared across one or more datasets. The...     

Ellipsometric measurements on tantalum and tantalum oxide films and variation of the optical constants with structure

The optical constants n and k for films of tantalum and naturally grown tantalum oxide were determined at wavelengths of 4960 Å, 5487 Å and 6049 Å using an ellipsometer at two angles of incidence. The tantalum films were prepared at a pressure equal to or less than 2 × 10^-8 Torr during evaporation. Values are compared with those of other workers. Optical observations on films with mixed b.c.c. and f.c.c. structures (as determined from X-ray measurements) indicate the presence of a b.c.c. and f.c.c. mixture throughout; this is contrary to previous suggestions by other workers that the f.c.c. phase was present in very thin films only and was converted to b.c.c. in thick films.     

In Vitro Simultaneous Transfer of Lipids to HDL in Coronary Artery Disease and in Statin Treatment

The exchange of lipids with cells and other lipoproteins is a crucial process in HDL metabolism and for HDL antiatherogenic function. Here, we tested a practical method to quantify the simultaneous transfer to HDL of phospholipids, free-cholesterol, esterified cholesterol and triacylglycerols and to verify the lipid transfer in patients with coronary artery disease (CAD) or undergoing statin treatment. Twenty-eight control subjects without CAD, 27 with CAD and 25 CAD patients under simvastatin treatment were studied. Plasma samples were incubated with a donor nanoemulsion prepared by ultrasonication of the constituent lipids and labeled with radioactive lipids; % lipids transferred to HDL were quantified in the HDL-containing supernatant after chemical precipitation of non-HDL fractions and the nanoemulsion. The assay was precise and reproducible. Increase of temperature (4–37 °C), of incubation period (5 min to 2 h), of HDL-cholesterol concentration (33–244 mg/dL) and of mass of nanoemulsion lipids (0.075–0.3 mg/μL) resulted in increased lipid transfer from the nanoemulsion to HDL. In contrast, increasing pH (6.5–8.5) and albumin concentration (3.5–7.0 g/dL) did not affect lipid transfer. There was no difference between CAD and control non-CAD with regard to the lipid transfer, but statin treatment reduced the transfer to HDL of all four lipids. The test herein described is a valid and practical tool for exploring an important aspect of HDL metabolism.     

The effect of chlorine and oxygen concentrations on the removal of mercury at an FGD-batch reactor

A series of laboratory scale experiments were conducted in an FGD-batch reactor. A synthetic flue gas was produced and directed through a CaCO₃ suspension contained in a glass reactor vessel. The suspension temperature was set at 54 °C through a water bath. In order to observe the distribution of mercury species in the system, solid, liquid and gaseous samples were taken and analysed. For gaseous mercury determination, continuous measurements were carried out, up and downstream the reactor. Furthermore, the concentration of chlorine in the scrubber solution of the system was varied from 0 to 62 g/l under different oxidative conditions.In a first approach, a concentration drop of elemental mercury coming out of the system was observed. The latter occurs only when high concentrations of Cl⁻ are present, combined with a high O₂ availability in the scrubber. It was also observed that mercury species distribution in the different phases varies, depending on the available chemical form of chlorine and oxygen concentration.     

Simultaneous determination of flavor enhancers in stock cube samples by using spectrophotometric data and multivariate calibration

Spectrophotometric data followed by a suitable treatment of chemometric analysis were used for the simultaneous determination of monosodium glutamate (MSG), guanosine 5′-monophosphate (GMP) and inosine 5′-monophosphate (IMP) in stock cube samples, without any previous extraction step. By this way, the overlapping of the absorption spectra was resolved using a PLS-1 model. The concentration for experimental calibration matrix were varied between 5.03–34.2 μg mL⁻¹ for IMP and GMP, and 448–1399 μg mL⁻¹ for MSG. The relative errors of prediction (REPCV %) were 1.8, 2.8 and 3.1 for IMP, GMP and MSG, respectively.     

An assessment of advanced mobile services acceptance: Contributions from TAM and diffusion theory models

Today, in addition to traditional mobile services, there are new ones already being used, thanks to the advances in 3G-related technologies. Our work contributed to the emerging body of research by integrating TAM and Diffusion Theory. Based on a sample of 542 Dutch consumers, we found that traditional antecedents of behavioral intention, ease of use and perceived usefulness, can be linked to diffusion-related variables, such as social influence and perceived benefits (flexibility and status).     

Emission inventory: An urban public policy instrument and benchmark

Global concern with climate change has led to the development of a variety of solutions to monitor and reduce emissions on both local and global scales. Under the United Nations Framework Convention on Climate Change (UNFCCC), both developed and emerging countries have assumed responsibility for developing and updating national inventories of greenhouse gas emissions from anthropic sources. This creates opportunities and incentives for cities to carry out their own local inventories and, thereby, develop air quality management plans including both essential key players and stakeholders at the local level. The aim of this paper is to discuss the role of local inventories as an urban public policy instrument and how this type of local instrument may bring advantages countrywide in enhancing the global position of a country. Local inventories have been carried out in many cities of the world and the main advantage of this is that it allows an overview of emissions produced by different municipal activities, thereby, helps decision makers in the elaboration of efficient air quality management plans. In that way, measures aimed at the reduction of fossil fuel consumption to lower local atmospheric pollution levels can also, in some ways, reduce GHG emissions.     

Hybrid integral transforms analysis of the bioheat equation with variable properties

Pennes’ equation is the most frequently employed model to describe heat transfer processes within living tissues, with numerous applications in clinical diagnostics and thermal treatments. A number of analytical solutions were provided in the literature that represent the temperature distribution across tissue structures, but considering simplifying assumptions such as uniform and linear thermophysical properties and blood perfusion rates. The present work thus advances such analysis path by considering a heterogeneous medium formulation that allows for spatially variable parameters across the tissue thickness. Besides, the eventual variation of blood perfusion rates with temperature is also accounted for in the proposed model. The Generalized Integral Transform Technique (GITT) is employed to yield a hybrid numerical–analytical solution of the bioheat model in heterogeneous media, which reduces to the exact solution obtained via the Classical Integral Transform Method for a linear formulation with uniform coefficients. The open source UNIT code (“UNified Integral Transforms”) is utilized to obtain numerical results for a set of typical values of the governing parameters, in order to illustrate the convergence behavior of the proposed eigenfunction expansions and inspect the importance of accounting for spatially variable properties in predicting the thermal response of living tissues to external stimulus.     

Wrapping up Metal–Organic Framework Crystals with Carbon Nanotubes

The presence of tetrazine units in the organic nodes of UiO-68-TZCD controls the formation of ultrathin coatings of single-wall nanotubes that decorate the surface of the crystal. These crystal hybrids can be prepared straightforwardly in one step and are extraordinarily respectful with the properties of the framework for combination of mesoporosity and surface areas ≈4.000 m² g⁻¹, with excellent stability in water, and conductivities at room temperature of 4 × 10⁻² S cm⁻¹ even at very low carbon weight contents (2.3 wt.%).