Can computer vision problems benefit from structured hierarchical classification?

Research in the field of supervised classification has mostly focused on the standard, so-called “flat” classification approach, where the problem classes live in a trivial, one-level semantic space. There is however an increasing interest in the hierarchical classification approach, where a performance gain is expected by incorporating prior taxonomic knowledge about the classes into the learning process. Intuitively, the hierarchical approach should be beneficial in general for the classification of visual content, as suggested by the fact that humans seem to organize objects into hierarchies based on visually perceived similarities. In this paper, we provide an analysis that aims to determine the conditions under which the hierarchical approach can consistently give better performances than the flat approach for the classification of visual content. In particular, we (1) show how hierarchical methods can fail to outperform flat methods when applied to real vision-based classification problems, and (2) investigate the underlying reasons for the lack of improvement, by applying the same methods to synthetic datasets in a simulation. Our conclusion is that the use of high-level hierarchical feature representations is crucial for obtaining a performance gain with the hierarchical approach, and that poorly chosen prior taxonomies hinder this gain even though proper high-level features are used.     

Trimetallic Mn‐Fe‐Ni Oxide Nanoparticles Supported on Multi‐Walled Carbon Nanotubes as High‐Performance Bifunctional ORR/OER Electrocatalyst in Alkaline Media

Discovering precious metal‐free electrocatalysts exhibiting high activity and stability toward both the oxygen reduction (ORR) and the oxygen evolution (OER) reactions remains one of the main challenges for the development of reversible oxygen electrodes in rechargeable metal–air batteries and reversible electrolyzer/fuel cell systems. Herein, a highly active OER catalyst, Fe_0.3Ni_0.7O_X supported on oxygen‐functionalized multi‐walled carbon nanotubes, is substantially activated into a bifunctional ORR/OER catalyst by means of additional incorporation of MnO_X. The carbon nanotube‐supported trimetallic (Mn‐Ni‐Fe) oxide catalyst achieves remarkably low ORR and OER overpotentials with a low reversible ORR/OER overvoltage of only 0.73 V, as well as selective reduction of O₂ predominantly to OH^?. It is shown by means of rotating disk electrode and rotating ring disk electrode voltammetry that the combination of earth‐abundant transition metal oxides leads to strong synergistic interactions modulating catalytic activity. The applicability of the prepared catalyst for reversible ORR/OER electrocatalysis is evaluated by means of a four‐electrode configuration cell assembly comprising an integrated two‐layer bifunctional ORR/OER electrode system with the individual layers dedicated for the ORR and the OER to prevent deactivation of the ORR activity as commonly observed in single‐layer bifunctional ORR/OER electrodes after OER polarization.     

Toward Second-Generation Cardiomyogenic and Anti-cardiofibrotic 1,4-Dihydropyridine-Class TGFβ Inhibitors

Innovative therapeutic modalities for pharmacological intervention of transforming growth factor β (TGFβ)-dependent diseases are of great value. b-Annelated 1,4-dihydropyridines (DHPs) might be such a class, as they induce TGFβ receptor type II degradation. However, intrinsic drawbacks are associated with this compound class and were systematically addressed in the presented study. It was possible to install polar functionalities and bioisosteric moieties at distinct sites of the molecules while maintaining TGFβ-inhibitory activities. The introduction of a 2-amino group or 7-N-alkyl modification proved to be successful strategies. Aqueous solubility was improved by up to seven-fold at pH 7.4 and 200-fold at pH 3 relative to the parent ethyl 4-(biphenyl-4-yl)-2,7,7-trimethyl-5-oxo-1,4,5,6,7,8-hexahydroquinoline-3-carboxylate. The therapeutic potential of the presented DHPs was further underscored in view of a potential dual mode of action: The differentiation of committed human iPSC-derived cardiac progenitor cells (CPCs) was potently stimulated, and the rescue of cardiac fibrosis phenotypes was observed in engineered heart tissue (EHT) constructs.     

Hot metal corrosion behavior for graphite refractory impregnated with TiO2, ZrO2 carrying solutions

This paper looked into hot metal corrosion behavior for graphite refractory used in blast furnace hearth, consisting of 50% graphite and 50% anthracite, impregnated with TiO₂, ZrO₂ carrying solutions. The method used for corrosion was the Finger Test, which was conducted in an induction furnace with 30 × 30 × 280 mm bar‐shaped test specimens, and the hot metal and slag used were taken from the CSN #2 Blast Furnace runner, the test temperature was 1470°C with a sixty‐minute isotherm. After the corrosion test, the specimens were characterized by their dimensional variation, X‐ray diffractometry and Scanning Electronic Microscopy (SEM).     

Influence of cracks on the local current–voltage parameters of silicon solar cells

The existence of cracks in silicon solar cells can drastically reduce the electrical performance of an individual cell and even of an entire photovoltaic module. An in‐depth understanding of the influence of cracks on solar cells enables therefore calculations of the crack impact and other following effects on module level. This paper shows a detailed analysis of the electrical influence of cracks with two different spatially resolved methods including global and local current–voltage characteristics. The main influence of cracks is an increased recombination current density in the depletion region, which is clearly shown by spatially resolved dark lock‐in thermography measurements with local current–voltage investigation. This increased recombination current density affects further cell parameters such as the efficiency, which is confirmed also by the global current–voltage characteristics. The additionally used ratio image technique based on electroluminescence measurements is in comparison with the local current–voltage method, the more reliable and faster method for the crack detection itself, and allows on cell‐level and module‐level a continuous inspection of cracks. Copyright © 2013 John Wiley & Sons, Ltd.     

A methodology to investigate the wear of blast furnace hearth carbon refractory lining

In this work it is presented a methodology applied by Companhia Siderúrgica Nacional (CSN, Brazil) in order to assess its blast furnace 3 carbon refractory hearth lining conditions. The aim of the investigation was the gathering of critical data base for blast furnace refractories life cycle follow up and eventual repair decision taking. The sample drilling locations were chosen around the tap holes and hearth bottom areas. The advancing drilling depths were based on local temperature profiles. The isotherm limit of 500 °C was elected as a control parameter to assess the critical carbon refractory condition. The guidelines for sampling and testing as well as the results obtained through the physical and chemical characterizations of the cored samples are presented. The condition of the refractory lining is discussed under the light of the different known wear mechanisms of blast furnace hearth carbon refractories.     

Opportunities (‘costs) matter: A comment on Pimentel and Patzek “Ethanol production using corn,switchgrass, and wood; biodiesel production using soybean and sunflower”

The energy balance for different crops reported by Pimentel and Patzek ignores opportunity costs. Including opportunity costs substantially changes the results and leads to different conclusions.     

Wind energy statistics for large arrays of wind turbines (New England and Central U.S. Regions)

The performance characteristics have been simulated for large dispersed arrays of 500–1500 kW wind turbines producing power and feeding it directly into the New England or Central U.S. utility distribution grids. These studies, based on design power performance curves, indicate that in good wind environments the 500 kW generators can average (on an annual basis) up to 240 kW mean power output, and the 1500 kW generators can average up to 350 kW mean power output. Higher mean power output (averaging up to 470 kW) is indicated, however from a hypothetical 1125 kW rated power unit designed to operate at wind speeds near those observed throughout the study area, rather than the higher design operating wind speed of the 1500 kW unit. The beneficial effect of operating large disperse arrays of wind turbines is that available power output can be increased—if winds are not blowing over one part of the array, chances are they will over some other part of the array. These studies indicate that wind power availability levels of 200 kW per 1125 kW generator were 77–93 per cent, depending on season. Reasonably steady high wind power in winter and high afternoon peak wind power in summer (corresponding to peak air conditioning load) means that significant peak load displacement can be achieved without the use of storage.     

Selection of Desulfurizing Agents and Optimization of Operational Variables in Hot Metal Desulfurization

The growing demand for steels with tighter compositional specifications led the Companhia Siderúrgica Nacional (CSN) to develop more efficient processes. To solve this problem this paper aims to identify the operational variables more impacting in the desulfurization process, specifically in torpedo car, as well as its causes and solutions. Then select and test, with laboratorial and industrial tests, desulfurizing agents based of CaC₂, CaO, CaCO₃, and Mg to assess the cost per quantity of product desulfurized. The mixture with best results was not that one with highest content of CaC₂. It is believed that this mixture showed better efficiency because of the increased agitation of the bath, produced by the releasing of gas from compound CaCO₃ present in this mixture.     

A simple hourly clear-sky solar radiation model based on meteorological parameters

An hourly solar radiation model, based on observed meteorological data, was developed and tested. As a means of comparison, the Watt and Bird models and the Solmet regression models were also tested. Several conclusions were drawn about the parameterization of solar radiation-depleting parameters. It was determined that a reasonable estimate of lower layer aerosol extinction can be determined using humidity, visibility, and mixing height. The parameterization of water vapor absorption obtained by atmospheric rather than laboratory observations was found to give better model results.