A self-adaptive synthetic over-sampling technique for imbalanced classification

Traditionally, in supervised machine learning, (a significant) part of the available data (usually 50%-80%) is used for training and the rest—for validation. In many problems, however, the data are highly imbalanced in regard to different classes or does not have good coverage of the feasible data space which, in turn, creates problems in validation and usage phase. In this paper, we propose a technique for synthesizing feasible and likely data to help balance the classes as well as to boost the performance in terms of confusion matrix as well as overall. The idea, in a nutshell, is to synthesize data samples in close vicinity to the actual data samples specifically for the less represented (minority) classes. This has also implications to the so-called fairness of machine learning. In this paper, we propose a specific method for synthesizing data in a way to balance the classes and boost the performance, especially of the minority classes. It is generic and can be applied to different base algorithms, for example, support vector machines, k-nearest neighbour classifiers deep neural, rule-based classifiers, decision trees, and so forth. The results demonstrated that (a) a significantly more balanced (and fair) classification results can be achieved and (b) that the overall performance as well as the performance per class measured by confusion matrix can be boosted. In addition, this approach can be very valuable for the cases when the number of actual available labelled data is small which itself is one of the problems of the contemporary machine learning.     

Sustainable ammonia production through process synthesis and global optimization

Current ammonia production technologies have a significant carbon footprint. In this study, we present a process synthesis and global optimization framework to discover the efficient utilization of renewable resources in ammonia production. Competing technologies are incorporated in a process superstructure where biomass, wind, and solar routes are compared with the natural gas-based reference case. A deterministic global optimization-based branch-and-bound algorithm is used to solve the resulting large-scale nonconvex mixed-integer nonlinear programming problem (MINLP). Case studies for Texas, California, and Iowa are conducted to examine the effects of different feedstock prices and availabilities. Results indicate that profitability of ammonia production is highly sensitive to feedstock and electricity prices, as well as greenhouse gas (GHG) restrictions. Under strict 75% GHG reductions, biomass to ammonia route is found to be competitive with natural gas route, whereas wind and solar to ammonia routes require further improvement to compete with those two routes. © 2018 American Institute of Chemical Engineers AIChE J, 65: e16498 2019     

Recent Biophysical Issues About the Preparation of Solute-Filled Lipid Vesicles

Here we report some recent biophysical issues on the preparation of solute-filled lipid vesicles and their relevance to the construction of “synthetic cells.” First, we introduce the “semi-synthetic minimal cells” as the liposome-based cell-like systems, which contain a minimal number of biomolecules required to display simple and complex biological functions. Next, we focus on recent aspects related to the construction of synthetic cells. Emphasis is given to the interplay between the methods of synthetic cell preparation and the physics of solute encapsulation. We briefly introduce the notion of structural and compositional “diversity” in synthetic cell populations.     

Optimization of waste quail eggshells as biocomposites for polyaniline in ammonia gas detection

A simple, cost-effective, and novel chemical sensor for ammonia (NH₃) gas detection was developed from polyaniline (PANI)/quail eggshell (QES) composites. QES is a natural waste enriched in calcium carbonate. In this work, pure PANI was synthesized from chemical oxidation method and PANI/QES composites were prepared from physical mixing of QES with the synthesized PANI at different mass ratio. A series of complementary techniques including Fourier transform infrared and ultraviolet-visible spectrometers, scanning electron microscope with energy dispersive detection coupled with mapping, thermogravimetric analysis, and X-ray diffractometer were used to characterize the physicochemical and textural properties of the biocomposites. From the results, PANI/QES composite with a mass ratio of 1 exhibited the lowest NH₃ detection limit of 5.24 ppm with a linear correlation coefficient (R²) of close to unity (0.9932) between the signal and NH₃ gas concentration. As a whole, the PANI/QES biocomposites synthesized from this work exhibited excellent selectivity toward NH₃ gas even in the presence of other gas impurities, such as acetone, ethanol, and hexane. For the sensor reusability, the PANI/QES biocomposites can be reused in the application of NH₃ gas detection for at least 4 cycles.     

Corrosion of hot-dip-galvanised steel and zinc alloy-coated steel in ammonia and ammonium chloride

There are many potential causes of corrosion in animal buildings. Animals exhale large quantities of moisture into the air creating high relative humidity in the building if the moisture is not properly vented. High humidity increases the potential for condensation. In addition, ammonia may be found in large quantities in animal buildings. Ammonia is released from manure and urine. In addition, ammonium chloride is used as a nitrogen source in fertilisers. In this study, the atmospheric corrosion of hot-dip-galvanised steel and zinc alloy-coated steel such as zinc–aluminium and zinc–aluminium–magnesium has been studied in atmospheres containing different levels of ammonia. Investigations have also been conducted at different levels of ammonium chloride. The results are discussed in view of the mechanisms of corrosion of zinc and zinc alloy-coated steel in ammonia and ammonium chloride-containing environments.     

High-resolution ISAR imaging of fast rotating targets based on pattern-coupled Bayesian strategy for multiple measurement vectors

Very high resolution inverse synthetic aperture radar (ISAR) imaging of fast rotating targets is a complicated task. There may be insufficient pulses or may introduce migration through range cells (MTRC) during the coherent processing interval (CPI) when we use the conventional range Doppler (RD) ISAR technique. With compressed sensing (CS) technique, we can achieve the high-resolution ISAR imaging of a target with limited number of pulses. Sparse representation based method can achieve the super resolution ISAR imaging of a target with a short CPI, during which the target rotates only a small angle and the range migration of the scatterers is small. However, traditional CS-based ISAR imaging method generally faced with the problem of basis mismatch, which may degrade the ISAR image. To achieve the high resolution ISAR imaging of fast rotating targets, this paper proposed a pattern-coupled sparse Bayesian learning method for multiple measurement vectors, i.e. the PC-MSBL algorithm. A multi-channel pattern-coupled hierarchical Gaussian prior is proposed to model the pattern dependencies among neighboring range cells and correct the MTRC problem. The expectation-maximization (EM) algorithm is used to infer the maximum a posterior (MAP) estimate of the hyperparameters. Simulation results validate the effectiveness and superiority of the proposed algorithm.     

Alteration of the Route to Menaquinone towards Isochorismate-Derived Metabolites

Chorismate and isochorismate constitute branch-point intermediates in the biosynthesis of many aromatic metabolites in microorganisms and plants. To obtain unnatural compounds, we modified the route to menaquinone in Escherichia coli. We propose a model for the binding of isochorismate to the active site of MenD ((1R,2S, 5S,6S)-2-succinyl-5-enolpyruvyl-6-hydroxycyclohex-3-ene-1-carboxylate (SEPHCHC) synthase) that explains the outcome of the native reaction with α-ketoglutarate. We have rationally designed variants of MenD for the conversion of several isochorismate analogues. The double-variant Asn117Arg–Leu478Thr preferentially converts (5S,6S)-5,6-dihydroxycyclohexa-1,3-diene-1-carboxylate (2,3-trans-CHD), the hydrolysis product of isochorismate, with a >70-fold higher ratio than that for the wild type. The single-variant Arg107Ile uses (5S,6S)-6-amino-5-hydroxycyclohexa-1,3-diene-1-carboxylate (2,3-trans-CHA) as substrate with >6-fold conversion compared to wild-type MenD. The novel compounds have been made accessible in vivo (up to 5.3 g L⁻¹). Unexpectedly, as the identified residues such as Arg107 are highly conserved (>94 %), some of the designed variations can be found in wild-type SEPHCHC synthases from other bacteria (Arg107Lys, 0.3 %). This raises the question for the possible natural occurrence of as yet unexplored branches of the shikimate pathway.     

Combining molybdenum carbide with ceria overlayers to boost Mo/CeO2 catalyzed ammonia synthesis

The design of an efficient non-noble metal catalyst is of burgeoning interest for ammonia synthesis. Herein, we report a Mo₂C/CeO₂ catalyst that is superior in ammonia synthesis activity. In this catalyst, molybdenum carbide coexisted with the ceria overlayers which is from the ceria support as the strong metal–support interaction. There is a high proportion of low-valent Mo species, as well as high concentration of Ce³⁺ and surface oxygen species. The presence of Mo₂C and CeO₂ overlayers not only leads to enhancement of hydrogen and nitrogen adsorption, but also facilitates the desorption and exchange of adsorbed species with the gaseous reagents. Compared with the Mo/CeO₂ catalyst prepared without carbonization, the Mo₂C/CeO₂ catalyst is more than sevenfold higher in ammonia synthesis rate. This work not only presents an explicit example of designing Mo-based catalyst that is highly efficient for ammonia synthesis by tuning the adsorption and desorption properties of the reactant gases, but opens a perspective for other elements in ammonia synthesis.     

Carbon supported bimetallic Ru‐Co catalysts for H2 production through NaBH4 and NH3BH3 hydrolysis

This work investigates the effect of the addition of small amounts of Ru (0.5‐1 wt%) to carbon supported Co (10 wt%) catalysts towards both NaBH₄ and NH₃BH₃ hydrolysis for H₂ production. In the sodium borohydride hydrolysis, the activity of Ru‐Co/carbon catalysts was sensibly higher than the sum of the activities of corresponding monometallic samples, whereas for the ammonia borane hydrolysis, the positive effect of Ru‐Co systems with regard to catalytic activity was less evident. The performances of Ru‐Co bimetallic catalysts correlated with the occurrence of an interaction between Ru and Co species resulting in the formation of smaller ruthenium and cobalt oxide particles with a more homogeneous dispersion on the carbon support. It was proposed that Ru^°, formed during the reduction step of the Ru‐Co catalysts, favors the H₂ activation, thus enhancing the reduction degree of the cobalt precursor and the number of Co nucleation centers. A subsequent reduction of cobalt and ruthenium species also occurs in the hydride reaction medium, and therefore the state of the catalyst before the catalytic experiment determines the state of the active phase formed in situ. The different relative reactivity of the Ru and Co active species towards the two investigated reactions accounted for the different behavior towards NaBH₄ and NH₃BH₃ hydrolysis.