Search

Generalizing segmentation networks that adapt to unseen target domains is practical to the real-world utility. Recent self-training-based approaches compute the pseudo labels of highly imbalanced target image that are typically biased to the majority classes and basically noisy. To avoid the model falling into an error-prone and suboptimal state, we propose an uncertainty-weighted prototype active learning method to deal with domain adaptation regarding the semantic segmentation task. For assisting active learning to select reliable samples, multiple prototypical anchors strategy is employed to build the target image distribution model and assign accurate pixel-level pseudo labels, which consists of two different stages. In the initial training stage, this model defines each class as a set of non-learnable prototypes, and performs online clustering to identify target instances for annotating in the prototype-based feature space. In the fine-tuning stage, entropy criterion is adopted to generate representative class prototypes set of target domain, then compute the uncertainty as the clustering weight of prototype-level features via a predictive entropy-based strategy. By mixing these latent representations of the target samples and developing feature alignment loss function, our method can achieve more accurate segmentation performance. Extensive experiments on two benchmarks, GTAV → Cityscapes and SYNTHIA → Cityscapes, demonstrate that the proposed model outperforms other state-of-the-art approaches, and ablation study verifies the effectiveness of each key component. We hope our work will open a new view for future research in the field of domain adaptation. The code will be released soon at https://github.com/zihaodong/UPAL. • Uncertainty-weighted prototype active learning assists domain adaptive segmentation. • Prototypical anchors strategy flexibly describes cross domain feature distribution. • Prototype-based selection solution considers both sample uncertainty and diversity. • Extensive experiments show the effectiveness of active domain adaptive segmentation. [ABSTRACT FROM AUTHOR]

Domain adaptive semantic segmentation is meaningful since collecting numerous labeled samples in different domains is expensive and time-consuming. Recent domain adaptation methods yield not so efficient performance compared with supervised learning. With the hypothesis that semantic feature can be shared across domains, this paper proposes a coarse-to-fine online matching architecture (COM) for one-stage domain adaptation. We consider subsequent learning stages progressively refining the task in the latent feature space, i.e. the finer set at each component is hierarchically derived from the coarser set of the previous components, including cross-domain global prototypes, categories and instances matching and anchor-points contrastive learning, which further achieve self-supervised learning with region-level pseudo label generated only in a single training step. Beforehand, feature refinement are performed to realize edge perception and inter-feature augmentation. Then, coarse-to-fine network fuses global and local consistency matching via specific distribution alignment between the source and target domain. Finally, the adversarial structure controls the uncertainty of generator prediction through the maximization of classification results and minimization of two classifiers discrepancy. This proposed method is evaluated in two unsupervised domain adaptation tasks, i.e. GTA5 → Cityscapes and SYNTHIA → Cityscapes. Extensive experiments verify the effectiveness of our proposed COM model and demonstrate its superiority over several state-of-the-art approaches. • Coarse-to-fine online matching fuses multi-level features for unified domain alignment. • Region-level self-supervised learning rectifies domain shift without extra parameter. • The predicted discrepancy between two classifiers denotes the uncertainty in network. • Experiments show the effectiveness of one-stage domain adaptive segmentation. [ABSTRACT FROM AUTHOR]

Because lithium resources are becoming increasingly scarce, their recovery is an urgent problem that needs to be solved. In this paper, an environmentally friendly process for the recovery of lithium ions from water resources by liquid-liquid extraction (LLE) is proposed. Menthol (DLM) was used as a hydrogen bond acceptor (HBA), and octanoic acid, decanoic acid, and lauric acid (LA) were used as hydrogen bond donors (HBDs) to form nine hydrophobic deep eutectic solvents (HDESs). The HDESs synthesized using DLM and LA had the best effect and thus were selected for the lithium extraction experiment. The single extraction efficiency of lithium ions was 80.69 %, with optimized rotational speed, equilibrium time, pH, molar ratio of organic phase to inorganic phase (O/L), and initial concentration of different lithium complexes of 660 rpm, 30 min, 12, 3:1, and 0.5 g/L, respectively. In addition, the regeneration cycle experiment of the HDES was further investigated. The possible extraction mechanism was demonstrated by Fourier-transform infrared spectroscopy and slope method. The separation factor (β) of Li/Na and Li/K obtained from the experimental results was 18.72 and 84.25, showing that 2DLM/LA can selectively extract lithium ions from systems containing Li, Na and K. This will be beneficial to design potential green HDESs for recovering lithium resources from aqueous solutions. • Nine hydrophobic deep eutectic solvents were selected for extraction experiment. • The effect of operating parameters on lithium ions was studied. • The single extraction efficiency of lithium was 80.69 % under optimal conditions. • A method for extracting lithium from the alkaline solution was provided. [ABSTRACT FROM AUTHOR]

BaTiO 3 terrace-like nano/microstructure were synthesized via a facile solvothermal method without surfacatant or template by controlling pH value. The phase structure, morphology and properties of the samples were characterized by XRD, FT-IR, XPS, SEM, UV–vis and ESR spectrum, etc. The piezoelectric potential analysis are calculated by finite element method simulation for BaTiO 3 prepared at different pH values under ultrasonic vibration. Morphology and crystal phase transform mechanism of BaTiO 3 at different pH values are proposed, respectively. The results show that pH value has a significant effect on the phase, crystallinity, morphology, size and piezocatalytic performance of BaTiO 3. When pH value is 10, BaTiO 3 exhibits spherical shape with an average size of ∼200 nm. With the increase of pH value, the morphology transforms grape-like shape from terrace-like shape. The transformation mechanism for BaTiO 3 is proposed to be a two–step growth model: initial nucleation and ripen, an oriented attachement accompanied by Ostwald ripening. H+ ions in OH− group occupy the Ba2+ position and affect the supersaturation of Ti (OH) 6 2−, thereby blocking the migration channel of Ba2+ and changing the crystal symmetry, resulting in changes in the content of tetragonal phase. FEM simulated shows the electric potential of BaTiO 3 is dominated by their morphologies. Compared with other samples, the best piezocatalytic performance of BaTiO 3 terrace-like nano/microstructure is demonstrated at pH = 12, which is the rate constant k of 15.5 × 10−3 min−1 by degrading 93 % of RhB dye solution in 180min, and keep almost unchanged after five cycles. The attractive piezocatalytic properties are considered to benefit from BaTiO 3 terrace-like nano/microstructure, which is composed of the numerous nanounits self-assembly micrometer-scale structure. [ABSTRACT FROM AUTHOR]

In this study, we are concerned with overcoming the limitation that most existing methods learn the dose prediction distribution without taking into account hard-to-predict regions. This paper presents a novel multi-stage dose prediction framework based on difficulty-aware learning to progressively predict 3D radiation dose distribution. We first stack multiple sub-models employing an encoder–decoder architecture to construct a multi-stage framework that decomposes the challenging prediction task into a few easy-to-learn sub-tasks for improving dose prediction. Moreover, a difficulty-aware learning strategy with voxel-level attention is developed to effectively handle hard-to-predict voxels or regions. Finally, the performance of our framework is compared to that of existing related models while being evaluated by voxel-based mean absolute error (MAE) and clinical-related dosimetric metrics. We evaluated the proposed method on the head-and-neck datasets from the 2020 AAPM OpenKBP challenge. Our framework achieved a dose score of 2.367 and a DVH score of 1.378, which outperformed the other existing single-stage and cascade dose prediction models. The average prediction error of the dose CI and dose HI in the test dataset of 56 Gy, 63 Gy, and 70 Gy were quite similar to the ground-truth plans (ranging from 0.01 to 0.02). For the OARs, the average error in predicting the D m a x value was 2.66 ± 2.88 Gy (brainstem), 1.94 ± 1.43 Gy (spinal cord), and 1.39 ± 1.67 Gy (mandible), whereas the values for the D m e a n value were 1.10 ± 1.07 Gy (left parotid), 1.13 ± 1.03 Gy (right parotid), 1.81 ± 1.77 Gy (esophagus), and 1.49 ± 1.66 Gy (larynx). Both quantitative and qualitative results demonstrated the potential of our framework and its superior performance compared to the state-of-the-art methods in the dose prediction task, which thus proved to be a promising tool for predicting the dose distribution in computer-aided radiotherapy planning. • A multi-stage framework to progressively predict 3D radiation dose. • Decompose the challenging prediction task into a few easy-to-learn sub-tasks. • A difficulty-aware mechanism to focus on the hard-to-predict regions. • Extensive experimental analysis indicates the effectiveness of our framework. [ABSTRACT FROM AUTHOR]

Abstract: In this study, highly ordered mesoporous silica material (SBA-15) functionalized with N-(quinoline-8-yl)-2-(3-triethoxysilyl-propylamino)acetamide (QTPA) as an INHIBIT (INH) logic gate was reported. The functionalized mesoporous SBA-15 silica (QTPA-SBA-15) was characterized by Powder X-ray diffraction (XRD), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FT-IR), 29Si solid state NMR spectra, N2 adsorption/desorption and thermal gravimetric analysis (TGA). The characterization data indicated that the organic ligand was successfully grafted into the mesopores of SBA-15. The fluorescent characterization revealed that this material showed excellent selectivity to Cu2+ over other cations even in trace amount. Moreover, due to the fluorophore moiety presented in the channels of SBA-15, QTPA-SBA-15 can be considered as a selective fluorescent probe for Cu2+. The fluorescent changes of QTPA-SBA-15 upon the addition of cations (Cu2+, Zn2+ and Cd2+) is utilized as an INH logic gate at the molecular level, using [Cu2+ and Zn2+] or [Cu2+ and Cd2+] as chemical inputs and the fluorescence intensity signal as output. [ABSTRACT FROM AUTHOR]

• Preparation of phosphate-based ILs and their application in separation of azeotropic. • Explored efficient green separation and recovery of methanol-containing azeotropes. • Revealed the molecular level separation mechanism of azeotropes. • Exploring the influence of ionic liquids on the relative volatility of azeotropes. Abundant coal resources in China have led to an excess of methanol production. This work explores the separation strategy of dimethoxymethane-methanol (DMM-MeOH) and dimethyl carbonate (DMC)-MeOH binary azeotropes, which are easily produced in the production of diesel additives. Implementing energy-saving, emission reduction, and low carbon emission strategies is important. Three types of phosphate esters ionic liquids (ILs) with different chain lengths were prepared to realize the separation of DMM-MeOH and DMC-MeOH binary azeotrope. Based on the vapor-liquid equilibrium experiments, the vapor-liquid equilibrium data of the DMM-MeOH and DMC-MeOH binary azeotropes using phosphate esters ILs of three different chain lengths were measured at 101.3 kPa. The effect of phosphate esters ILs with three different chain lengths on the separation of DMM-MeOH and DMC-MeOH azeotropes was investigated. The separation mechanism of the azeotropes was studied at the molecular level. The electron density difference explores the optimal configuration between molecules. The atomic theory of molecules analysis directly displays the characteristics of bond critical points and explores the influence of hydrogen bond strength on the intermolecular bond length. The important roles of hydrogen bonds and steric hindrance in the separation process were investigated by reduced density gradient and independent gradient model. Quantum chemical calculation results showed that phosphate esters ILs have a significant effect on the molecular interactions of azeotropes, which is the main factor for phosphate esters ILs in reducing the relative volatility between azeotropes. [ABSTRACT FROM AUTHOR]

[Display omitted] • Quaternary ammonium salt can efficiently remove nitrides from oil by forming DESs. • The extraction properties of four quaternary ammonium salts were compared. • The mechanism of extraction process was studied by quantum chemical calculation. • The high recovery efficiency of quaternary ammonium salt was studied. Nitrogen compounds significantly affect the quality of oil products; therefore, the removal of these compounds from oil products has attracted significant attention. In this study, a novel method for separating nitrogen compounds from oil using quaternary ammonium salt (QAS) as an extractant to form deep eutectic solvents (DESs) with non-basic nitrides, interacting with basic nitrides was proposed. Based on the COSMO-SAC model, the possibility of nitrogen compounds acting as hydrogen-bond donors in oil was verified. The effects of four QAS types, agent oil ratios, reaction temperatures, and stirring times on the removal of nitrogen compounds in simulated oil, and the effects of the stripping agent type and recovery times on the recovery rate of QAS were investigated. The DESs were analyzed and characterized by Fourier-transform infrared spectroscopy. The results indicated that tetraethyl ammonium chloride (TEAC) was the best denitrogenation agent. When the stirring time was 8 min, the removal rate of the nitrogen compounds reached 92.62%. Simultaneously, the removal mechanism was studied at the molecular level using a quantum chemical calculation method. Ethyl acetate was used as the stripping agent. After five recovery cycles, the recovery rate of TEAC was 91.5%. This method demonstrates a high selectivity, rapid extraction efficiency, and high recovery efficiency. This study provides theoretical guidance for the efficient and green removal of nitrogen compounds from oil products. [ABSTRACT FROM AUTHOR]

[Display omitted] • Two microbial consortia with high lignocellulolytic capability were enriched. • Combined with LAB, these consortia facilitated biodegradability of rice straw. • Bioaugmentation improved mono- and disaccharide contents and enzymatic hydrolysis. • Synergistic action of microbial consortia and LAB enhanced methane production. Two consortia of lignocellulolytic microbes (CL and YL) were isolated from the rumen of ruminants. Their ability to facilitate the degradation of rice straw and enhance methane (CH 4) production were evaluated, both individually and combined with lactic acid bacteria (LAB). After 30 days of degradation, rice straw powders (RSPs) were observed to change in physical structure and also displayed a significant reduction in lignocellulose content. Combined application of microbial consortia with LAB efficiently improved enzymatic hydrolysis of RSPs, increasing organic acid as well as mono- and disaccharide contents. Synergistic action between microbial consortia and LAB enhanced CH 4 yield, and rice straw treated with YL + LAB had the highest CH 4 production (357.53 mL CH 4 /g VS), more than fivefold of the control. The newly identified microbial consortia are capable of efficiently degrading lignocellulosic biomass. Functioning synergistically with LAB, they provide a feasible way biodegrade rice straw and enhance methane production from agricultural wastes. [ABSTRACT FROM AUTHOR]