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.     

Low-cost automated capillary electrophoresis instrument assembled from commercially available parts

A CE instrument that can be assembled from commercially available components with minimal construction effort is described. Except for the electronic control circuitry no specially made parts are required. It is based on a flexible design of microfluidic, electropneumatic, and electronic sections and different configurations can easily be implemented. Automated injection into the capillary is performed hydrodynamically by the application of a pressure for a controlled length of time. The performance of the device was tested with a contactless conductivity detector by separating different metal ions. In addition, nine metal cations related to the quality of honey were separated in 2.3 min and four honey samples were analysed quantitatively to demonstrate the applicability of the method.     

Unbalanced optimal transport: Dynamic and Kantorovich formulations

This article presents a new class of distances between arbitrary nonnegative Radon measures inspired by optimal transport. These distances are defined by two equivalent alternative formulations: (i) a dynamic formulation defining the distance as a geodesic distance over the space of measures (ii) a static “Kantorovich” formulation where the distance is the minimum of an optimization problem over pairs of couplings describing the transfer (transport, creation and destruction) of mass between two measures. Both formulations are convex optimization problems, and the ability to switch from one to the other depending on the targeted application is a crucial property of our models. Of particular interest is the Wasserstein–Fisher–Rao metric recently introduced independently by [7], [15]. Defined initially through a dynamic formulation, it belongs to this class of metrics and hence automatically benefits from a static Kantorovich formulation.     

Supramolecular Tuning Enables Selective Oxygen Reduction Catalyzed by Cobalt Porphyrins for Direct Electrosynthesis of Hydrogen Peroxide

We report a supramolecular strategy for promoting the selective reduction of O₂ for direct electrosynthesis of H₂O₂. We utilized cobalt tetraphenylporphyrin (Co-TPP), an oxygen reduction reaction (ORR) catalyst with highly variable product selectivity, as a building block to assemble the permanently porous supramolecular cage Co-PB-1(6) bearing six Co-TPP subunits connected through twenty-four imine bonds. Reduction of these imine linkers to amines yields the more flexible cage Co-rPB-1(6). Both Co-PB-1(6) and Co-rPB-1(6) cages produce 90–100 % H₂O₂ from electrochemical ORR catalysis in neutral pH water, whereas the Co-TPP monomer gives a 50 % mixture of H₂O₂ and H₂O. Bimolecular pathways have been implicated in facilitating H₂O formation, therefore, we attribute this high H₂O₂ selectivity to site isolation of the discrete molecular units in each supramolecule. The ability to control reaction selectivity in supramolecular structures beyond traditional host–guest interactions offers new opportunities for designing such architectures for a broader range of catalytic applications.     

Local linear convergence analysis of Primal–Dual splitting methods

In this paper, we study the local linear convergence properties of a versatile class of Primal–Dual splitting methods for minimizing composite non-smooth convex optimization problems. Under the assumption that the non-smooth components of the problem are partly smooth relative to smooth manifolds, we present a unified local convergence analysis framework for these methods. More precisely, in our framework, we first show that (i) the sequences generated by Primal–Dual splitting methods identify a pair of primal and dual smooth manifolds in a finite number of iterations, and then (ii) enter a local linear convergence regime, which is characterized based on the structure of the underlying active smooth manifolds. We also show how our results for Primal–Dual splitting can be specialized to cover existing ones on Forward–Backward splitting and Douglas–Rachford splitting/ADMM (alternating direction methods of multipliers). Moreover, based on these obtained local convergence analysis result, several practical acceleration techniques are discussed. To exemplify the usefulness of the obtained result, we consider several concrete numerical experiments arising from fields including signal/image processing, inverse problems and machine learning. The demonstration not only verifies the local linear convergence behaviour of Primal–Dual splitting methods, but also the insights on how to accelerate them in practice.     

Carbon Dots as a New Class of Diamagnetic Chemical Exchange Saturation Transfer (diaCEST) MRI Contrast Agents

While carbon dots (C‐dots) have been extensively investigated pertaining to their fluorescent, phosphorescent, electrochemiluminescent, optoelectronic, and catalytic features, their inherent chemical exchange saturation transfer magnetic resonance imaging (CEST MRI) properties are unknown. By virtue of their hydrophilicity and abundant exchangeable protons of hydroxyl, amine, and amide anchored on the surface, we report here that C‐dots can be adapted as effective diamagnetic CEST (diaCEST) MRI contrast agents. As a proof‐of‐concept demonstration, human glioma cells were labeled with liposomes with or without encapsulated C‐dots and implanted in mouse brain. In vivo CEST MRI was able to clearly differentiate labeled cells from non‐labeled cells. The present findings may encourage new applications of C‐dots for in vivo imaging in deep tissues, which is currently not possible using conventional fluorescent (near‐infrared) C‐dots.     

Versatile transamination in quinonediimine chemistry: Towards a novel class of water soluble UV/violet chromophores

The transamination reaction of 2,5-diaminobenzoquinonediimine (QDI) with ethylenediamine gave fluorescent 1,2,3,4-tetrahydropyrazino[2,3-g]quinoxaline (1). When the same reaction was carried out with N,N’-bis(aminoethyl)-1,3-propanediamine, a novel cationic quinoxalinium species (2) was isolated, which can be further condensed with p-cyanobenzaldehyde to afford a benzimidazolo-fused quinoxaline dye (3) that is a water-soluble fluorophore in the UV–visible range.     

Validating additive correction schemes against gradient-based extrapolations

The use of additive correction schemes to obtain structures and vibrational frequencies of increasingly larger molecules is becoming more common. Such approaches, based on the cubic extrapolation formula applied directly to the quantity of interest, have been successfully validated only at the highest levels of computational accuracy: for coupled cluster methods with comparably large basis sets. Here, a systematic validation of geometries and vibrational frequencies is carried out, including more affordable and relevant levels of theory, such as the Møller-Plesset perturbation theory applied with smaller basis sets. Comparisons of such additive schemes against the more rigorous gradient-based extrapolation are presented. The cbs () routine of the open-source quantum-chemistry package Psi4 has been extended for this purpose. The results confirm that geometries and frequencies of covalently bound species obtained with additive correction schemes are in an excellent agreement with the results of gradient-based extrapolations. However, when applied to systems involving noncovalent interactions, the errors due to such schemes are significantly larger. In general, we propose the application of gradient-based extrapolations, as they incur no extra cost compared to additive schemes.