Co-Training for Visual Object Recognition Based on Self-Supervised Models Using a Cross-Entropy Regularization

Automatic recognition of visual objects using a deep learning approach has been successfully applied to multiple areas. However, deep learning techniques require a large amount of labeled data, which is usually expensive to obtain. An alternative is to use semi-supervised models, such as co-training, where multiple complementary views are combined using a small amount of labeled data. A simple way to associate views to visual objects is through the application of a degree of rotation or a type of filter. In this work, we propose a co-training model for visual object recognition using deep neural networks by adding layers of self-supervised neural networks as intermediate inputs to the views, where the views are diversified through the cross-entropy regularization of their outputs. Since the model merges the concepts of co-training and self-supervised learning by considering the differentiation of outputs, we called it Differential Self-Supervised Co-Training (DSSCo-Training). This paper presents some experiments using the DSSCo-Training model to well-known image datasets such as MNIST, CIFAR-100, and SVHN. The results indicate that the proposed model is competitive with the state-of-art models and shows an average relative improvement of 5% in accuracy for several datasets, despite its greater simplicity with respect to more recent approaches.     

Ternary choline chloride/caffeic acid/ethylene glycol deep eutectic solvent as both a monomer and template in a molecularly imprinted polymer

A molecularly imprinted polymer based on a ternary deep eutectic solvent comprised of choline chloride/caffeic acid/ethylene glycol was prepared. The caffeic acid in the ternary deep eutectic solvent was used as both a monomer and template. The molecularly imprinted polymer based on the ternary deep eutectic solvent was characterized by Fourier transform infrared spectroscopy, thermogravimetric analysis, field‐emission scanning electron microscopy, Brunauer–Emmett–Teller surface area analysis, atomic force microscopy, and elemental analysis. A series of molecularly imprinted polymers based on choline chloride/caffeic acid/ethylene glycol with different molar ratios was prepared and applied to the molecular recognition of polyphenols. A comparison of the recognition ability of molecularly imprinted polymers to polyphenols revealed that the choline chloride/caffeic acid/ethylene glycol (1:0.4:1, molar ratio) molecularly imprinted polymer had the best molecular recognition effect with 132 μg/g of protocatechuic acid, 104 μg/g of catechins, 80 μg/g of epicatechin, and 123 μg/g of caffeic acid in 6 h, as well as good molecular recognition ability for polyphenols from a Radix Asteris sample. These results show that the ternary deep eutectic solvent based molecularly imprinted polymer is a potential medium that can be applied to drug purification, drug delivery, and drug analysis.     

Pipette‐tip solid‐phase extraction based on deep eutectic solvent modified graphene for the determination of sulfamerazine in river water

A green and novel deep eutectic solvent modified graphene was prepared and used as a neutral adsorbent for the rapid determination of sulfamerazine in a river water sample by pipette‐tip solid‐phase extraction. Compared with conventional graphene, deep eutectic solvent modified graphene can change the surface of graphene with wrinkled structure and higher selective extraction ability. The properties of deep eutectic solvent modified graphene and graphene were characterized by scanning electron microscopy, Fourier transform infrared spectroscopy, and thermogravimetric analysis. Static adsorption showed deep eutectic solvent modified graphene had a higher adsorption ability (18.62 mg/g) than graphene. Under the optimum conditions, factors such as kinds of washing solvents and elution solvents and volume of elution solvent were evaluated. The limits of detection and quantification were 0.01 and 0.03 μg/mL, respectively. The method recoveries of sulfamerazine were in the range of 91.01–96.82% with associated intraday relative standard deviations ranging from 1.63 to 3.46% and interday relative standard deviations ranging from 0.68 to 3.84%. Deep eutectic solvent modified graphene showed satisfactory results (recovery was 95.38%) and potential for rapid purification of sulfamerazine in river water sample in combination with the pipette‐tip solid‐phase extraction method.     

A Sustainable Redox-Flow Battery with an Aluminum-Based,Deep-Eutectic-Solvent Anolyte

Nonaqueous redox-flow batteries are an emerging energy storage technology for grid storage systems, but the development of anolytes has lagged far behind that of catholytes due to the major limitations of the redox species, which exhibit relatively low solubility and inadequate redox potentials. Herein, an aluminum-based deep-eutectic-solvent is investigated as an anolyte for redox-flow batteries. The aluminum-based deep-eutectic solvent demonstrated a significantly enhanced concentration of circa 3.2 m in the anolyte and a relatively low redox potential of 2.2 V vs. Li⁺/Li. The electrochemical measurements highlight that a reversible volumetric capacity of 145 Ah L⁻¹ and an energy density of 189 Wh L⁻¹ or 165 Wh kg⁻¹ have been achieved when coupled with a I₃⁻/I⁻ catholyte. The prototype cell has also been extended to the use of a Br₂-based catholyte, exhibiting a higher cell voltage with a theoretical energy density of over 200 Wh L⁻¹. The synergy of highly abundant, dendrite-free, multi-electron-reaction aluminum anodes and environmentally benign deep-eutectic-solvent anolytes reveals great potential towards cost-effective, sustainable redox-flow batteries.     

Exploration of a ternary deep eutectic solvent of methyltriphenylphosphonium bromide/chalcone/formic acid for the selective recognition of rutin and quercetin in Herba Artemisiae Scopariae

Methyltriphenylphosphonium bromide/chalcone/formic acid, a green ternary deep eutectic solvent, was applied as a functional monomer and dummy template simultaneously in the synthesis of a new molecularly imprinted polymer. Ternary deep eutectic solvent based molecularly imprinted polymers are used as a solid‐phase extraction sorbent in the separation and purification of rutin and quercetin from Herba Artemisiae Scopariae combined with high‐performance liquid chromatography. Fourier transform infrared spectroscopy and field‐emission scanning electron microscopy were applied to characterize the deep eutectic solvent based molecularly imprinted polymers synthesized using different molar ratios of chalcone. The static and competitive adsorption tests were performed to examine the recognition ability of the molecularly imprinted polymers to rutin and quercetin. The ternary deep eutectic solvent consisting of formic acid/chalcone/methyltriphenylphosphonium bromide (1:0.05:0.5) had the best molecular recognition effect. After optimization of the washing solvents (methanol/water, 1:9) and eluting solvents (acetonitrile/acetic acid, 9:1), a reliable analytical method was developed for strong recognition towards rutin and quercetin in Herba Artemisiae Scopariae with satisfactory extraction recoveries (rutin: 92.48%, quercetin: 94.23%). Overall, the chalcone ternary deep eutectic solvent‐based molecularly imprinted polymer coupled with solid‐phase extraction is an effective method for the selective purification of multiple bioactive compounds in complex samples.     

基于模糊灰色新陈代谢FGM(1,1)模型预测及应用探讨

巷道围岩变形是煤矿开采普遍存在的问题,巷道开挖过程中围岩会出现破裂现象.主要探讨模糊灰色系统的深部巷道围岩变形预测模型及应用,通过分析模糊数学理论、灰色理论,给出模糊灰色系统预测新陈代谢模型.     

促进深度学习的专题复习教学*——以“碳及其化合物”为例

以促进学生的深度学习为基本理念,围绕“自然界的碳循环”“大气中CO₂的控减排”“化学家们合成的新型碳家族成员”等板块,设计并成功实施系列情境问题,在系列问题的讨论、分析、解决过程中帮助学生建构“碳”家族成员间相互转化的知识网络,提高学生思维的纵深度,引发高阶思维,促进深度学习。     

One-pot rapid synthesis of 4H-1-benzopyran derivatives in a deep eutectic solvent

A method for introducing a biologically active heterocycle, 2-methylquinoline into the 4-position of a 2-amino-4H-1-benzopyran skeleton is described. Choline chloride/glucose (1:1 molar ratio) was used as both the solvent and catalyst in the reaction of a salicylaldehyde, methylquinoline, and cyanoacetate to obtain 2-amino-4H-1-benzopyran derivatives in 48%–80% yields after short reaction times. The effects of the deep eutectic solvent type, substrate molar ratio, cosolvent, temperature, and reaction time were examined. The method has the advantages of simple steps, environmental friendliness, mild conditions, and wide substrate applicability. This is the first attempt to synthesize methylquinoline derivatives of 4H-1-benzopyran.     

Microwave-Assisted One-Pot Lipid Extraction and Glycolipid Production from Oleaginous Yeast Saitozyma podzolica in Sugar Alcohol-Based Media

Glycolipids are non-ionic surfactants occurring in numerous products of daily life. Due to their surface-activity, emulsifying properties, and foaming abilities, they can be applied in food, cosmetics, and pharmaceuticals. Enzymatic synthesis of glycolipids based on carbohydrates and free fatty acids or esters is often catalyzed using certain acyltransferases in reaction media of low water activity, e.g., organic solvents or notably Deep Eutectic Systems (DESs). Existing reports describing integrated processes for glycolipid production from renewables use many reaction steps, therefore this study aims at simplifying the procedure. By using microwave dielectric heating, DESs preparation was first accelerated considerably. A comparative study revealed a preparation time on average 16-fold faster than the conventional heating method in an incubator. Furthermore, lipids from robust oleaginous yeast biomass were successfully extracted up to 70% without using the pre-treatment method for cell disruption, limiting logically the energy input necessary for such process. Acidified DESs consisting of either xylitol or sorbitol and choline chloride mediated the one-pot process, allowing subsequent conversion of the lipids into mono-acylated palmitate, oleate, linoleate, and stearate sugar alcohol esters. Thus, we show strong evidence that addition of immobilized Candida antarctica Lipase B (Novozym 435^®), in acidified DES mixture, enables a simplified and fast glycolipid synthesis using directly oleaginous yeast biomass.     

Solubility and Stability of Some Pharmaceuticals in Natural Deep Eutectic Solvents-Based Formulations

Some medicines are poorly soluble in water. For tube feeding and parenteral administration, liquid formulations are required. The discovery of natural deep eutectic solvents (NADES) opened the way to potential applications for liquid drug formulations. NADES consists of a mixture of two or more simple natural products such as sugars, amino acids, organic acids, choline/betaine, and poly-alcohols in certain molar ratios. A series of NADES with a water content of 0–30% (w/w) was screened for the ability to solubilize (in a stable way) some poorly water-soluble pharmaceuticals at a concentration of 5 mg/mL. The results showed that NADES selectively dissolved the tested drugs. Some mixtures of choline-based NADES, acid-neutral or sugars-based NADES could dissolve chloral hydrate (dissociated in water), ranitidine·HCl (polymorphism), and methylphenidate (water insoluble), at a concentration of up to 250 mg/mL, the highest concentration tested. Whereas a mixture of lactic-acid–propyleneglycol could dissolve spironolacton and trimethoprim at a concentration up to 50 and 100 mg/mL, respectively. The results showed that NADES are promising solvents for formulation of poorly water-soluble medicines for the development of parenteral and tube feeding administration of non-water-soluble medicines. The chemical stability and bioavailability of these drug in NADES needs further studies.