Tooth Type Enhanced Transformer for Children Caries Diagnosis on Dental Panoramic Radiographs.

The objective of this study was to introduce a novel deep learning technique for more accurate children caries diagnosis on dental panoramic radiographs. Specifically, a swin transformer is introduced, which is compared with the state-of-the-art convolutional neural network (CNN) methods that are widely used for caries diagnosis. A tooth type enhanced swin transformer is further proposed by considering the differences among canine, molar and incisor. Modeling the above differences in swin transformer, the proposed method was expected to mine domain knowledge for more accurate caries diagnosis. To test the proposed method, a children panoramic radiograph database was built and labeled with a total of 6028 teeth. Swin transformer shows better diagnosis performance compared with typical CNN methods, which indicates the usefulness of this new technique for children caries diagnosis on panoramic radiographs. Furthermore, the proposed tooth type enhanced swin transformer outperforms the naive swin transformer with the accuracy, precision, recall, F1 and area-under-the-curve being 0.8557, 0.8832, 0.8317, 0.8567 and 0.9223, respectively. This indicates that the transformer model can be further improved with a consideration of domain knowledge instead of a copy of previous transformer models designed for natural images. Finally, we compare the proposed tooth type enhanced swin transformer with two attending doctors. The proposed method shows higher caries diagnosis accuracy for the first and second primary molars, which may assist dentists in caries diagnosis.

Deep Learning for Free-Hand Sketch: A Survey.

Free-hand sketches are highly illustrative, and have been widely used by humans to depict objects or stories from ancient times to the present. The recent prevalence of touchscreen devices has made sketch creation a much easier task than ever and consequently made sketch-oriented applications increasingly popular. The progress of deep learning has immensely benefited free-hand sketch research and applications. This paper presents a comprehensive survey of the deep learning techniques oriented at free-hand sketch data, and the applications that they enable. The main contents of this survey include: (i) A discussion of the intrinsic traits and unique challenges of free-hand sketch, to highlight the essential differences between sketch data and other data modalities, e.g., natural photos. (ii) A review of the developments of free-hand sketch research in the deep learning era, by surveying existing datasets, research topics, and the state-of-the-art methods through a detailed taxonomy and experimental evaluation. (iii) Promotion of future work via a discussion of bottlenecks, open problems, and potential research directions for the community.

Context Aware Convolutional Neural Network for Children Caries Diagnosis on Dental Panoramic Radiographs.

The objective of this study is to improve traditional convolutional neural networks for more accurate children dental caries diagnosis on panoramic radiographs. A context aware convolutional neural network (CNN) is proposed by considering information among adjacent teeth, based on the fact that caries of teeth often affects each other due to the same growing environment. Specifically, when performing caries diagnosis on a tooth, information from its adjacent teeth will be collected and adaptively fused for final classification. Children panoramic radiographs of 210 patients with one or more caries and 94 patients without caries are utilized, among which there are a total of 6028 teeth with 3039 to be caries. The proposed context aware CNN outperforms typical CNN baseline with the accuracy, precision, recall, F1 score, and area-under-the-curve (AUC) being 0.8272, 0.8538, 0.8770, 0.8652, and 0.9005, respectively, showing potential to improve typical CNN instead of just copying them in previous works. Specially, the proposed method performs better than two five-year attending doctors for the second primary molar caries diagnosis. Considering the results obtained, it is beneficial to promote CNN based deep learning methods for assisting dentists for caries diagnosis in hospitals.

Learning Macromanagement in Starcraft by Deep Reinforcement Learning.

StarCraft is a real-time strategy game that provides a complex environment for AI research. Macromanagement, i.e., selecting appropriate units to build depending on the current state, is one of the most important problems in this game. To reduce the requirements for expert knowledge and enhance the coordination of the systematic bot, we select reinforcement learning (RL) to tackle the problem of macromanagement. We propose a novel deep RL method, Mean Asynchronous Advantage Actor-Critic (MA3C), which computes the approximate expected policy gradient instead of the gradient of sampled action to reduce the variance of the gradient, and encode the history queue with recurrent neural network to tackle the problem of imperfect information. The experimental results show that MA3C achieves a very high rate of winning, approximately 90%, against the weaker opponents and it improves the win rate about 30% against the stronger opponents. We also propose a novel method to visualize and interpret the policy learned by MA3C. Combined with the visualized results and the snapshots of games, we find that the learned macromanagement not only adapts to the game rules and the policy of the opponent bot, but also cooperates well with the other modules of MA3C-Bot.

Top management team faultlines, green technology innovation and firm financial performance.

Focusing on green technology innovation, the paper researches its motivation and implementation results from the perspective of internal corporate management. The impact of top manager team faultlines on green technology innovation and the moderating role of environmental information disclosure in the relationship between green technology innovation and financial performance are discussed. Using listed manufacturing corporate data obtained from China Stock Market & Accounting Research database, the hypotheses are tested. It is found that there is an inverted U-shaped relationship between task-related faultline and green technology innovation, while bio-demographic faultline has no significant influence on green technology innovation. In addition, both green technology innovation and environmental information disclosure have positive effects on financial performance, while the moderating effect of environmental information disclosure on innovation-performance doesn't exist. Combined with the research results, the practical implications for policy makers and managers and future research avenue are proposed.

Atomic-scale phase separation induced clustering of solute atoms.

Dealloying typically occurs via the chemical dissolution of an alloy component through a corrosion process. In contrast, here we report an atomic-scale nonchemical dealloying process that results in the clustering of solute atoms. We show that the disparity in the adatom-substrate exchange barriers separate Cu adatoms from a Cu-Au mixture, leaving behind a fluid phase enriched with Au adatoms that subsequently aggregate into supported clusters. Using dynamic, atomic-scale electron microscopy observations and theoretical modeling, we delineate the atomic-scale mechanisms associated with the nucleation, rotation and amorphization-crystallization oscillations of the Au clusters. We expect broader applicability of the results because the phase separation process is dictated by the inherent asymmetric adatom-substrate exchange barriers for separating dissimilar atoms in multicomponent materials.

Environmental Management Breadth, Environmental Management Depth, and Manufacturing Performance.

According to the number of environmental management practices implemented by the firm and the degree of integration of environmental management with other functions of the firm, this study divides environmental management into two dimensions: environmental management breadth and environmental management depth. We argue that the impact of environmental management breadth on manufacturing performance is moderated by environmental management depth. A survey data including 225 Chinese manufacturing firms is used to test the hypotheses. Results show that there is an inverted U-shaped relationship between environmental management breadth and manufacturing performance; the impact of environmental management depth is positive; meanwhile, environmental management depth moderates the relationship between environmental management breadth and manufacturing performance.

Influence of Environmental Management on Green Process Innovation: Comparison of Multiple Mediating Effects Based on Routine Replication.

Although green process technology is vital to sustainable development, few articles focus on how to implement it from the perspective of firms. This article tries to answer this question. Being set as an antecedent of green process innovation, the influence of environmental management is analyzed and the influential path is elaborated. Hypotheses are tested by means of multivariate regression analysis and bootstrap method. The results show that environmental management is conducive to firms' green process innovation, and the influence is through zero-order routine replication and higher-order routine replication. The mediating effect played by the interaction between the two is stronger than that of the individual. Implications are given to academia and practitioners.

Nanocrystal Conversion-Assisted Design of Sn-Fe Alloy with a Core-Shell Structure as High-Performance Anodes for Lithium-Ion Batteries.

Sn-based alloy materials are strong candidates to replace graphitic carbon as the anode for the next generation lithium-ion batteries because of their much higher gravimetric and volumetric capacity. A series of nanosize Sn y Fe alloys derived from the chemical transformation of preformed Sn nanoparticles as templates have been synthesized and characterized. An optimized Sn5Fe/Sn2Fe anode with a core-shell structure delivered 541 mAh·g-1 after 200 cycles at the C/2 rate, retaining close to 100% of the initial capacity. Its volumetric capacity is double that of commercial graphitic carbon. It also has an excellent rate performance, delivering 94.8, 84.3, 72.1, and 58.2% of the 0.1 C capacity (679.8 mAh/g) at 0.2, 0.5, 1 and 2 C, respectively. The capacity is recovered upon lowering the rate. The exceptional cycling/rate capability and higher gravimetric/volumetric capacity make the Sn y Fe alloy a potential candidate as the anode in lithium-ion batteries. The understanding of Sn y Fe alloys from this work also provides insight for designing other Sn-M (M = Co, Ni, Cu, Mn, etc.) system.

The Influence of Top Managers on Environmental Information Disclosure: The Moderating Effect of Company's Environmental Performance.

Abundant extant literature emphasizes the impact of board members attributes' influence on environmental information disclosure. Considering the voluntary nature of environmental information disclosure, executives have strong managerial discretion when they make such decisions, so this article focuses on top managers' influence on environmental information disclosure. We hypothesize that top managers' educational background and age will affect companies' environmental decision. The hypotheses are verified with the data from Chinese listed manufacturing companies. As the results show, a Master of Business Administration (MBA) educational background and average age of top managers positively affect environmental information disclosure, while the impact of legal educational background is negative. The company's environmental performance plays a U-shaped moderating effect on the relationship between MBA educational background and environmental information disclosure.

Multiview Clustering via Unified and View-Specific Embeddings Learning.

Multiview clustering, which aims at using multiple distinct feature sets to boost clustering performance, has a wide range of applications. A subspace-based approach, a type of widely used methods, learns unified embedding from multiple sources of information and gives a relatively good performance. However, these methods usually ignore data similarity rankings; for example, example A may be more similar to B than C, and such similarity triplets may be more effective in revealing the data cluster structure. Motivated by recent embedding methods for modeling knowledge graph in natural-language processing, this paper proposes to mimic different views as different relations in a knowledge graph for unified and view-specific embedding learning. Moreover, in real applications, it happens so often that some views suffer from missing information, leading to incomplete multiview data. Under such a scenario, the performance of conventional multiview clustering degenerates notably, whereas the method we propose here can be naturally extended for incomplete multiview clustering, which enables full use of examples with incomplete feature sets for model promotion. Finally, we demonstrate through extensive experiments that our method performs better than the state-of-the-art clustering methods.

In situ atomic-scale observation of inhomogeneous oxide reduction.

We report in situ atomic-scale transmission electron microscopy observations of the surface dynamics during Cu2O reduction. We show inhomogeneous oxide reduction caused by the preferential adsorption of hydrogen at step edges that induces oxygen loss and destabilizes Cu atoms within the step edge, thereby resulting in the retraction motion of atomic steps at the oxide surface.

Electrochemical Performance of Nanosized Disordered LiVOPO4.

ε-LiVOPO4 is a promising multielectron cathode material for Li-ion batteries that can accommodate two electrons per vanadium, leading to higher energy densities. However, poor electronic conductivity and low lithium ion diffusivity currently result in low rate capability and poor cycle life. To enhance the electrochemical performance of ε-LiVOPO4, in this work, we optimized its solid-state synthesis route using in situ synchrotron X-ray diffraction and applied a combination of high-energy ball-milling with electronically and ionically conductive coatings aiming to improve bulk and surface Li diffusion. We show that high-energy ball-milling, while reducing the particle size also introduces structural disorder, as evidenced by 7Li and 31P NMR and X-ray absorption spectroscopy. We also show that a combination of electronically and ionically conductive coatings helps to utilize close to theoretical capacity for ε-LiVOPO4 at C/50 (1 C = 153 mA h g-1) and to enhance rate performance and capacity retention. The optimized ε-LiVOPO4/Li3VO4/acetylene black composite yields the high cycling capacity of 250 mA h g-1 at C/5 for over 70 cycles.

Dislocation nucleation facilitated by atomic segregation.

Surface segregation-the enrichment of one element at the surface, relative to the bulk-is ubiquitous to multi-component materials. Using the example of a Cu-Au solid solution, we demonstrate that compositional variations induced by surface segregation are accompanied by misfit strain and the formation of dislocations in the subsurface region via a surface diffusion and trapping process. The resulting chemically ordered surface regions acts as an effective barrier that inhibits subsequent dislocation annihilation at free surfaces. Using dynamic, atomic-scale resolution electron microscopy observations and theory modelling, we show that the dislocations are highly active, and we delineate the specific atomic-scale mechanisms associated with their nucleation, glide, climb, and annihilation at elevated temperatures. These observations provide mechanistic detail of how dislocations nucleate and migrate at heterointerfaces in dissimilar-material systems.

Phase diagram and structural evolution of tin/indium (Sn/In) nanosolder particles: from a non-equilibrium state to an equilibrium state.

A binary system of tin/indium (Sn/In) in the form of nanoparticles was investigated for phase transitions and structural evolution at different temperatures and compositions. The Sn/In nanosolder particles in the composition range of 24-72 wt% In were synthesized by a surfactant-assisted chemical reduction method under ambient conditions. The morphology and microstructure of the as-synthesized nanoparticles were analyzed by scanning electron microscopy (SEM), high resolution transmission electron microscopy (HRTEM), selected area electron diffraction (SAED) and X-ray diffraction (XRD). HRTEM and SAED identified InSn4 and In, with some Sn being detected by XRD, but no In3Sn was observed. The differential scanning calorimetry (DSC) thermographs of the as-synthesized nanoparticles exhibited an endothermic peak at around 116 °C, which is indicative of the metastable eutectic melting of InSn4 and In. When the nanosolders were subjected to heat treatment at 50-225 °C, the equilibrium phase In3Sn appeared while Sn disappeared. The equilibrium state was effectively attained at 225 °C. A Tammann plot of the DSC data of the as-synthesized nanoparticles indicated that the metastable eutectic composition is about 62% In, while that of the DSC data of the 225 °C heat-treated nanoparticles yielded a eutectic composition of 54% In, which confirmed the attainment of the equilibrium state at 225 °C. The phase boundaries estimated from the DSC data of heat-treated Sn/In nanosolder particles matched well with those in the established Sn-In equilibrium phase diagram. The phase transition behavior of Sn/In nanosolders leads to a new understanding of binary alloy particles at the nanoscale, and provides important information for their low temperature soldering processing and applications.

Atomic Structural Evolution during the Reduction of α-Fe2O3 Nanowires.

The atomic-scale reduction mechanism of α-Fe2O3 nanowires by H2 was followed using transmission electron microscopy to reveal the evolution of atomic structures and the associated transformation pathways for different iron oxides. The reduction commences with the generation of oxygen vacancies that order onto every 10th [Formula: see text] plane. This vacancy ordering is followed by an allotropic transformation of α-Fe2O3 → γ-Fe2O3 along with the formation of Fe3O4 nanoparticles on the surface of the γ-Fe2O3 nanowire by a topotactic transformation process, which shows 3D correspondence between the structures of the product and its host. These observations demonstrate that the partial reduction of α-Fe2O3 nanowires results in the formation of a unique hierarchical structure of hybrid oxides consisting of the parent oxide phase, γ-Fe2O3, as the one-dimensional wire and the Fe3O4 in the form of nanoparticles decorated on the parent oxide skeleton. We show that the proposed mechanism is consistent with previously published and our density functional theory results on the thermodynamics of surface termination and oxygen vacancy formation in α-Fe2O3. Compared to previous reports of α-Fe2O3 directly transformed to Fe3O4, our work provides a more in-depth understanding with substeps of reduction, i.e., the whole reduction process follows: α-Fe2O3 → α-Fe2O3 superlattice → γ-Fe2O3 + Fe3O4→ Fe3O4.

What Happens to LiMnPO4 upon Chemical Delithiation?

Olivine MnPO4 is the delithiated phase of the lithium-ion-battery cathode (positive electrode) material LiMnPO4, which is formed at the end of charge. This phase is metastable under ambient conditions and can only be produced by delithiation of LiMnPO4. We have revealed the manganese dissolution phenomenon during chemical delithiation of LiMnPO4, which causes amorphization of olivine MnPO4. The properties of crystalline MnPO4 obtained from carbon-coated LiMnPO4 and of the amorphous product resulting from delithiation of pure LiMnPO4 were studied and compared. The phosphorus-rich amorphous phases in the latter are considered to be MnHP2O7 and MnH2P2O7 from NMR, X-ray absorption spectroscopy, and X-ray photoelectron spectroscopy analysis. The thermal stability of MnPO4 is significantly higher under high vacuum than at ambient condition, which is shown to be related to surface water removal.

Cross-Modal Subspace Learning via Pairwise Constraints.

In multimedia applications, the text and image components in a web document form a pairwise constraint that potentially indicates the same semantic concept. This paper studies cross-modal learning via the pairwise constraint and aims to find the common structure hidden in different modalities. We first propose a compound regularization framework to address the pairwise constraint, which can be used as a general platform for developing cross-modal algorithms. For unsupervised learning, we propose a multi-modal subspace clustering method to learn a common structure for different modalities. For supervised learning, to reduce the semantic gap and the outliers in pairwise constraints, we propose a cross-modal matching method based on compound ℓ21 regularization. Extensive experiments demonstrate the benefits of joint text and image modeling with semantically induced pairwise constraints, and they show that the proposed cross-modal methods can further reduce the semantic gap between different modalities and improve the clustering/matching accuracy.

Robust Subspace Clustering With Complex Noise.

Subspace clustering has important and wide applications in computer vision and pattern recognition. It is a challenging task to learn low-dimensional subspace structures due to complex noise existing in high-dimensional data. Complex noise has much more complex statistical structures, and is neither Gaussian nor Laplacian noise. Recent subspace clustering methods usually assume a sparse representation of the errors incurred by noise and correct these errors iteratively. However, large corruptions incurred by complex noise cannot be well addressed by these methods. A novel optimization model for robust subspace clustering is proposed in this paper. Its objective function mainly includes two parts. The first part aims to achieve a sparse representation of each high-dimensional data point with other data points. The second part aims to maximize the correntropy between a given data point and its low-dimensional representation with other points. Correntropy is a robust measure so that the influence of large corruptions on subspace clustering can be greatly suppressed. An extension of pairwise link constraints is also proposed as prior information to deal with complex noise. Half-quadratic minimization is provided as an efficient solution to the proposed robust subspace clustering formulations. Experimental results on three commonly used data sets show that our method outperforms state-of-the-art subspace clustering methods.

In situ visualization of metallurgical reactions in nanoscale Cu/Sn diffusion couples.

The Cu-Sn metallurgical soldering reaction in two-segmented Cu-Sn nanowires is studied by in situ transmission electron microscopy. By varying the relative lengths of Cu and Sn segments, we show that the metallurgical reaction results in a Cu-Sn solid solution for small Sn/Cu length ratio while Cu-Sn intermetallic compounds (IMCs) for larger Sn/Cu length ratios. Upon heating the nanowires to ∼500 °C, two phase transformation pathways occur, η-Cu6Sn5 → ε-Cu3Sn → δ-Cu41Sn11 for nanowires with a long Cu segment and η-Cu6Sn5 → ε-Cu3Sn → γ-Cu3Sn with a short Cu segment. The evolution of Kirkendall voids in the nanowires demonstrates that Cu diffuses faster than Sn in IMCs. Void growth results in the nanowire breakage that shuts off the inter-diffusion of Cu and Sn and thus leads to changes in the phase transformation pathway in the IMCs.