Conditional Deep Gaussian Processes: Empirical Bayes Hyperdata Learning

It is desirable to combine the expressive power of deep learning with Gaussian Process (GP) in one expressive Bayesian learning model. Deep kernel learning showed success as a deep network used for feature extraction. Then, a GP was used as the function model. Recently, it was suggested that, albeit training with marginal likelihood, the deterministic nature of a feature extractor might lead to overfitting, and replacement with a Bayesian network seemed to cure it. Here, we propose the conditional deep Gaussian process (DGP) in which the intermediate GPs in hierarchical composition are supported by the hyperdata and the exposed GP remains zero mean. Motivated by the inducing points in sparse GP, the hyperdata also play the role of function supports, but are hyperparameters rather than random variables. It follows our previous moment matching approach to approximate the marginal prior for conditional DGP with a GP carrying an effective kernel. Thus, as in empirical Bayes, the hyperdata are learned by optimizing the approximate marginal likelihood which implicitly depends on the hyperdata via the kernel. We show the equivalence with the deep kernel learning in the limit of dense hyperdata in latent space. However, the conditional DGP and the corresponding approximate inference enjoy the benefit of being more Bayesian than deep kernel learning. Preliminary extrapolation results demonstrate expressive power from the depth of hierarchy by exploiting the exact covariance and hyperdata learning, in comparison with GP kernel composition, DGP variational inference and deep kernel learning. We also address the non-Gaussian aspect of our model as well as way of upgrading to a full Bayes inference.     

A Study on Railway Surface Defects Detection Based on Machine Vision

The detection of rail surface defects is an important tool to ensure the safe operation of rail transit. Due to the complex diversity of track surface defect features and the small size of the defect area, it is difficult to obtain satisfying detection results by traditional machine vision methods. The existing deep learning-based methods have the problems of large model sizes, excessive parameters, low accuracy and slow speed. Therefore, this paper proposes a new method based on an improved YOLOv4 (You Only Look Once, YOLO) for railway surface defect detection. In this method, MobileNetv3 is used as the backbone network of YOLOv4 to extract image features, and at the same time, deep separable convolution is applied on the PANet layer in YOLOv4, which realizes the lightweight network and real-time detection of the railway surface. The test results show that, compared with YOLOv4, the study can reduce the amount of the parameters by 78.04%, speed up the detection by 10.36 frames per second and decrease the model volume by 78%. Compared with other methods, the proposed method can achieve a higher detection accuracy, making it suitable for the fast and accurate detection of railway surface defects.     

DBS-Based Eutectogels: Organized Vessels to Perform the Michael Addition Reaction**

Supramolecular eutectogels were obtained from the gelation of 1,3 : 2,4-dibenzylidene-D-sorbitol (DBS) in cholinium chloride-based deep eutectic solvents (DES), differing for the nature of the hydrogen bond donor. Ethylene glycol, diethylene glycol, triethylene glycol, glycerol and urea were tested. Soft materials were fully characterized, determining critical gelation concentration, gel-sol transition temperatures and mechanical properties. Furthermore, to have information about the organization of the gelator in the tridimensional network, resonance light scattering, circular dichroism and microscopy investigations were performed. Eutectogels were used as organized “vessels” to perform the L-proline catalyzed Michael addition reaction. The probe reaction was carried out in gel phase and in DES solution. Data collected shed light on the effect that gel microenvironment exerts on the outcome of the reaction. In general, gel phases allowed having comparable or even better results than the ones collected in DES solution, with better results obtained in soft materials with the highest organization, as accounted for by the presence of larger aggregates and the occurrence of stronger intermolecular interactions. In turn, this accounts also for the effect of substrates structure that indicates that better yields could be obtained in the presence of more flexible nucleophile and dienones, having more extended π-surface.     

Rapid extraction of bioactive compounds from gardenia fruit using new and recyclable deep eutectic solvents

The utilization of deep eutectic solvent as an alternative and environmentally friendly option has gained significant attention. This study first proposed a series of benzylammonium chloride based-deep eutectic systems for the extraction of bioactive compounds from Gardenia jasminoides Ellis. Through the implementation of response surface methodology, the optimal solvent was determined to be dodecyldimethylbenzylammonium chloride–levulinic acid (1:3, mol/mol) with 35% (v/v) water, specifically tailored to extract geniposide, genipin-1-β-d -gentiobioside, crocin-1, and crocin-2 from gardenia fruits with the ratio of solid to liquid of 1:20 at 86°C for 16 min. Their total extraction yields could reach 70.6 mg/g, outperforming those obtained by other solvents and corresponding techniques. Furthermore, the eutectic system was retrieved after first-cycle extraction, and then applied in the subsequent extraction progress, yielding a consistent extraction efficiency of 97.1%. As compared to previous traditional methods, a quick, high-yielding, and green extraction procedure was achieved through simple heating settings that did not constrain the instrument. Therefore, dodecyldimethylbenzylammonium chloride–levulinic acid could serve as a sustainable and reusable solvent for efficient extraction of natural bioactive compounds from plant-based raw materials. The application of deep eutectic solvents has demonstrated their potential as designable solvents with stronger extraction capabilities than traditional organic solvents.     

Embedded palladium nanoparticles on metal–organic framework/covalently sulfonated magnetic SBA-15 mesoporous silica composite: As a highly proficient nanocatalyst for Suzuki–Miyaura coupling reaction in amino acid-based natural deep eutectic solvent

In this project, the main aim is the design and present a novel and unique heterogeneous nanocatalyst based on a metal–organic framework/covalently sulfonated magnetic SBA-15 mesoporous silica composite with the emphasis on promoting clean and green synthetic transformations and increasing the catalytic properties. In more detail, initially, SBA-15 containing magnetic nanoparticles was functionalized by a 1,3-propane sultone ligand. In the next step, the functionalized mesoporous substrate was used as a scaffold for the growth and synthesis of the zeolite imidazolate framework-8 (ZIF-8) crystals. The obtained composite was further applied as a suitable support for the immobilization of Pd nanoparticles via a post-modification procedure and the generation of heterogeneous catalysts. The prepared Fe₃O₄@SBA-15-SO₃H@ZIF-8@Pd was incorporated as a heterogeneous and green catalyst in the Suzuki coupling reaction in the natural deep eutectic solvent with efficient recyclability.     

Detection of small amounts of H2O in CO2‐rich fluid inclusions using Raman spectroscopy

Raman spectroscopic analysis at low (−100 °C) or high (100–200 °C) temperature is shown to be effective for detecting small amounts of H₂O in CO₂‐rich fluid inclusions from the deep lithosphere, which have previously been thought to be water‐free. Copyright © 2009 John Wiley & Sons, Ltd.     

Water Induced Alterations in Self-Assembly of a Bio-Surfactant in Deep Eutectic Solvent for Enhanced Enzyme Activity

Deep eutectic solvents (DESs) meet important requirements for green solvent technology, including non-toxicity, biodegradability, sustainability, and affordability. Despite possessing low cohesive energy density than water, DESs have been found to support the self-assembly of amphiphiles. It is very much pertinent to examine the effect of water on self-assembly of surfactants in DESs as the presence of water alters the inherent structure of DES, which is expected to affect the characteristic properties of self-assembly. Following this, we have investigated the self-assembly of amino-acid based surfactant, Sodium N-lauroyl sarcosinate (SLS), in DES-water mixtures (10, 30 and 50 w/w% of water) and explored the catalytic activity of Cytochrome-c (Cyt-c) in the formed colloidal systems. Investigations using surface tension, fluorescence, dynamic light scattering (DLS), and isothermal titration calorimetry (ITC) have shown that DES-water mixtures promote the aggregation of SLS, resulting in the lower critical aggregation concentration (cac ∼1.5–6-fold) of the surfactant as compared to water. The nanoclustering of DES at low water content and it's complete de-structuring at high water content affects the self-assembly in a contrasting manner governed by different set of interactions. Further, Cyt-c dispersed in DES-water colloidal solutions demonstrated 5-fold higher peroxidase activity than that observed in phosphate buffer.     

化学学历案促进深度学习的发生——化学平衡移动复习课

学历案是关于学习经历或过程的方案，利用学历案构建一个“以学为主、以教促学”的教学模式，改变了教师的主导地位，凸显了学生的主体地位。笔者尝试借助学历案，以花青素为载体，通过生活化的化学实验，将化学平衡常数K作为分析工具，进行证据推理，以外界条件使花青素溶液颜色变化，到花青素的工业合成为明线；以化学平衡移动的本质到化学平衡移动原理的应用为暗线，双线结合，学生自主建构“化学平衡移动”的认知模型，从而促进深度学习的发生。     

Gas-Free Amino- and Alkoxycarbonylation of Aryl Iodides in a Bioinspired Deep Eutectic Solvent with Mo(CO)6 as a Safe CO Source

The Pd-catalyzed amino- and alkoxycarbonylation of aryl iodides has been exploited, for the first time, in a bioinspired Deep Eutectic Solvent and under gas-free conditions, by using Mo(CO)₆ as the CO source. The method allows for the preparation of carboxylic amides and esters in high yields (up to 99 %), short reaction time (2 h) and under mild reaction conditions (80 °C), with a low catalyst loading (2.5 mol %). Noteworthy, in the case of N-hexylbenzamide, it has been demonstrated that both the catalyst and DES can be used for four consecutive runs, with a moderate decrease of catalytic efficiency. The methodology has been also applied to the preparation of an Active Pharmaceutical Ingredient used for the treatment of human scabies and lice.     

基于核心素养发展的高中化学深度教学——探究配合物的性质

基于深度教学理论，从主体性、对话性和关联性，广度、深度和参与度等开展“配合物性质探究”的深度教学。以层层推进的探究实验和科学严谨的推理论证，建立配合物性质和结构的相互关系，搭建配合物电离平衡模型，逐步实现从宏观到微观，从定性到定量的高阶思维培养。在探讨问题的过程中，形成学科认识思路，发展化学学科核心素养。