Recent deep learning based methods have achieved the state-of-the-art performance for handwritten Chinese character recognition (HCCR) by learning discriminative representations directly from raw ...data. Nevertheless, we believe that the long-and-well investigated domain-specific knowledge should still help to boost the performance of HCCR. By integrating the traditional normalization-cooperated direction-decomposed feature map (directMap) with the deep convolutional neural network (convNet), we are able to obtain new highest accuracies for both online and offline HCCR on the ICDAR-2013 competition database. With this new framework, we can eliminate the needs for data augmentation and model ensemble, which are widely used in other systems to achieve their best results. This makes our framework to be efficient and effective for both training and testing. Furthermore, although directMap+convNet can achieve the best results and surpass human-level performance, we show that writer adaptation in this case is still effective. A new adaptation layer is proposed to reduce the mismatch between training and test data on a particular source layer. The adaptation process can be efficiently and effectively implemented in an unsupervised manner. By adding the adaptation layer into the pre-trained convNet, it can adapt to the new handwriting styles of particular writers, and the recognition accuracy can be further improved consistently and significantly. This paper gives an overview and comparison of recent deep learning based approaches for HCCR, and also sets new benchmarks for both online and offline HCCR.
•Comprehensive study of handwritten Chinese character recognition (HCCR).•New benchmarks for both online and offline HCCR under a general framework.•Combination of convNet and domain-specific knowledge of directMap.•Writer adaptation of deep convolutional neural networks.•State-of-the-art performance on the ICDAR-2013 competition database.
•In non-differential case, the counterexamples of Jumarie’s basic fractional calculus formulas are given .•In theory, we showed that those counterexamples for differentiable functions also hold for ...non-differentiable cases.•We discuss these counterexamples from different points of view.
Jumarie proposed a modified Riemann–Liouville derivative definition and gave three so-called basic fractional calculus formulae such as Leibniz rule (u(t)v(t))(α)=u(α)(t)v(t)+u(t)v(α)(t), where u and v are required to be non-differentiable and continuous at the point t. We once gave the counterexamples to show that Jumarie’s formulae are not true for differentiable functions. In the paper, we give further counterexamples to prove that in non-differentiable cases these Jumarie’s formulae are also not true. Therefore, we proved that Jumarie’s formulae are not true for both cases of differentiable and non-differentiable functions, and then those results on fractional soliton equations obtained by using Jumarie’s formulae are not right.
Two new and similar cobalt(III) complexes, CoL2·NO3 (1) and CoL2·Cl (2), where L is 5-bromo-2-((2-(phenylamino)ethylimino)methyl)phenolate, have been synthesized and characterized by IR and UV-Vis ...spectra. Structures of the complexes were confirmed by single crystal X-ray determination. The Co atoms in the complexes are in octahedral coordination, with the donor atoms come from the two Schiff base ligands, viz. phenolate oxygen, and imino and amino nitrogen. The anions of the cobalt salts crystallized as counteranions in the complexes. The complexes were assayed for antibacterial activities by MTT method.
•A unified fire-induced concrete spalling theory is proposed.•Fire-induced spalling is categorised into three types according to their distinct governing mechanisms, viz., thermo-hygral, ...thermo-mechanical and thermo-chemical spalling.•The spalling temperature range, influencing factors and the protective measures for each type of spalling are discussed.•A design concept of ‘multiple defense line against fire-induced concrete spalling’ is introduced based on the unified fire-induced concrete spalling theory.
Spalling of concrete is a great potential threat to fire resistance of concrete structures. Understanding the underlying mechanism is important to predict and mitigate this unfavorable phenomenon. Currently, there are two main mechanisms to explain the fire-induced concrete spalling: viz. spalling due to (a) pore pressure buildup or (b) thermal stress. The relative importance of these two mechanisms has been a subject of intense debate in the research community over the past few decades. This paper presents a critical review of conflicting and concordant points on concrete spalling at high temperature and proposes a unified and coherent fire-induced concrete spalling theory. Therein, the authors propose three types of thermal spalling depending on the governing mechanisms: thermo-hygral spalling, thermo-mechanical spalling and thermo-chemical spalling. The criteria to forecast each type of spalling are established and the spalling temperature range for each of them is analysed. The spalling pattern, influencing factors and preventive measures for each type of spalling are also discussed in this paper.
Summary
Arbuscular mycorrhizal (AM) symbiosis relies on the formation of arbuscules for efficient nutrient exchange between plants and AM fungi. In this study, we identified a novel kinase gene in ...rice named OsADK1 (Arbuscule Development Kinase 1) that is required for arbuscule development.
By obtaining OsADK1pro::GUS transgenic rice plants and also generating Osadk1 mutants via CRISPR/Cas9 technique, OsADK1 was revealed to be specifically induced in the arbusculated cortical cells and mutations in OsADK1 resulted in an extremely low colonisation rate (c. 3%) of rice roots by AM fungus Rhizophagus irregularis.
In the mutant roots, the very few observed arbuscules nearly all arrested at an early ‘trunk‐forming’ phase without forming any branches. Increasing the inoculum strength of AM fungus or cocultivation with a wild‐type nurse plant did not result in the rescue of the arbuscule phenotype. Transcriptome sequencing of both nursed and un‐nursed Osadk1 mutants then revealed that the mutation of OsADK1 could greatly affect the AM symbiotic programme, including many key transcription factors such as RAM1 and WRI5.
OsADK1 therefore represents a new rice kinase that is required for arbuscule branching. Its identification opens a new window to explore the elaborate signal transduction pathway that determines arbuscule development during plant–fungus symbiosis.
π‐Conjugated organic materials possess a wide range of tunable optoelectronic properties which are dictated by their molecular structure and supramolecular arrangement. While many efforts have been ...put into tuning the molecular structure to achieve the desired properties, rational supramolecular control remains a challenge. Here, we report a novel series of supramolecular materials formed by the co‐assembly of weak π‐electron donor (indolo2,3‐acarbazole) and acceptor (aromatic o‐quinones) molecules via complementary hydrogen bonding. The resulting polarization creates a drastic perturbation of the molecular energy levels, causing strong charge transfer in the weak donor–acceptor pairs. This leads to a significant lowering (up to 1.5 eV) of the band gaps, intense absorption in the near‐IR region, very short π‐stacking distances (≥3.15 Å), and strong ESR signals in the co‐crystals. By varying the strength of the acceptor, the characteristics of the complexes can be tuned between intrinsic, gate‐, or light‐induced semiconductivity with a p‐type or ambipolar transport mechanism.
Supramolecular materials formed by the co‐assembly of weak π‐electron donor (indolo2,3‐acarbazole) and acceptor (aromatic o‐quinones) molecules are reported. Hydrogen bonding causes a strong charge transfer in the weak donor–acceptor pairs. By varying the strength of the acceptor, the characteristics of the complexes can be tuned between intrinsic, gate‐, or light‐induced semiconductivity with a p‐type or ambipolar transport mechanism.
Red luminescence is found in off‐white tris(iodoperchlorophenyl)methane (3I‐PTMH) crystals which is characterized by a high photoluminescence quantum yield (PLQY 91 %) and color purity (CIE ...coordinates 0.66, 0.34). The emission originates from the doublet excited state of the neutral radical 3I‐PTMR, which is spontaneously formed and becomes embedded in the 3I‐PTMH matrix. The radical defect can also be deliberately introduced into 3I‐PTMH crystals which maintain a high PLQY with up to 4 % radical concentration. The immobilized iodinated radical demonstrates excellent photostability (estimated half‐life >1 year under continuous irradiation) and intriguing luminescent lifetime (69 ns). TD‐DFT calculations demonstrate that electron‐donating iodine atoms accelerate the radiative transition while the rigid halogen‐bonded matrix suppresses the nonradiative decay.
Red luminescence in off‐white tris(iodoperchlorophenyl)methane crystals is characterized by a high quantum yield (PLQY 91 %) and color purity. The emission originates from the doublet excited state of the neutral radical defect. TD‐DFT calculations demonstrate that electron‐donating iodine atoms accelerate the radiative transition while the rigid halogen‐bonded matrix suppresses the nonradiative decay.
Multi-oriented and multi-lingual scene text detection plays an important role in computer vision area and is challenging due to the wide variety of text and background. In this paper, first, we point ...out the two key tasks when extending convolutional neural network (CNN)-based object detection frameworks to scene text detection. The first task is to localize the text region by a downsampled segmentation-based module, and the second task is to regress the boundaries of text region determined by the first task. Second, we propose a scene text detection framework based on fully convolutional network with a bi-task prediction module, in which one is a pixel-wise classification between the text and non-text and the other is pixel-wise regression to determine the vertex coordinates of quadrilateral text boundaries. Post-processing for word-level detection is based on non-maximum suppression, and for the line-level detection, we design a heuristic line segments grouping method to localize long text lines. We evaluated the proposed framework on various benchmarks, including multi-oriented and multi-lingual scene text data sets, and achieved the state-of-the-art performance on most of them. We also provide abundant ablation experiments to analyze several key factors in building high performance CNN-based scene text detection systems.
Renal tubules are the major component of the kidney and are vulnerable to a variety of injuries including hypoxia, proteinuria, toxins, metabolic disorders, and senescence. It has long been believed ...that tubules are the victim of injury. In this review, we shift this concept to renal tubules as a driving force in the progression of kidney diseases. In response to injury, tubular epithelial cells undergo changes and function as inflammatory and fibrogenic cells, with the consequent production of various bioactive molecules that drive interstitial inflammation and fibrosis. Innate immune-sensing receptors on the tubular epithelium also aggravate immune responses. Necroinflammation, an autoamplification loop between tubular cell death and interstitial inflammation, leads to the exacerbation of renal injury. Furthermore, tubular cells also play an active role in progressive renal injury via emerging mechanisms associated with a partial epithelial-mesenchymal transition, cell-cycle arrest at both G1/S and G2/M check points, and metabolic disorder. Thus, a better understanding the mechanisms by which tubular injury drives inflammation and fibrosis is necessary for the development of therapeutics to halt the progression of chronic kidney disease.
Silicene, a silicon analogue of graphene, has attracted increasing attention during the past few years. As early as in 1994, the possibility of stage corrugation in the Si analogs of graphite had ...already been theoretically explored. But there were very few studies on silicene until 2009, when silicene with a low buckled structure was confirmed to be dynamically stable by ab initio calculations. In spite of the low buckled geometry, silicene shares most of the outstanding electronic properties of planar graphene (e.g., the "Dirac cone", high Fermi velocity and carrier mobility). Compared with graphene, silicene has several prominent advantages: (1) a much stronger spin-orbit coupling, which may lead to a realization of quantum spin Hall effect in the experimentally accessible temperature, (2) a better tunability of the band gap, which is necessary for an effective field effect transistor (FET) operating at room temperature, (3) an easier valley polarization and more suitability for valleytronics study. From 2012, monolayer silicene sheets of different superstructures were successfully synthesized on various substrates, including Ag(111), Ir(111), ZrB2(0001), ZrC(111) and MoS2 surfaces. Multilayer silicene sheets have also been grown on Ag(111) surface. The experimental successes have stimulated many efforts to explore the intrinsic properties as well as potential device applications of silicene, including quantum spin Hall effect, quantum anomalous Hall effect, quantum valley Hall effect, superconductivity, band engineering, magnetism, thermoelectric effect, gas sensor, tunneling FET, spin filter, and spin FET, etc. Recently, a silicene FET has been fabricated, which shows the expected ambipolar Dirac charge transport and paves the way towards silicene-based nanoelectronics. This comprehensive review covers all the important theoretical and experimental advances on silicene to date, from the basic theory of intrinsic properties, experimental synthesis and characterization, modulation of physical properties by modifications, and finally to device explorations.
PDF
