Semantic guided knowledge graph for large-scale zero-shot learning

Zero-shot learning has received growing attention, which aims to improve generalization to unseen concepts. The key challenge in zero-shot tasks is to precisely model the relationship between seen and unseen classes. Most existing zero-shot learning methods capture inter-class relationships via a shared embedding space, leading to inadequate use of relationships and poor performance. Recently, knowledge graph-based methods have emerged as a new trend of zero-shot learning. These methods use a knowledge graph to accurately model the inter-class relationships. However, the currently dominant method for zero-shot learning directly extracts the fixed connection from off-the-shelf WordNet, which will inherit the inherent noise in WordNet. In this paper, we propose a novel method that adopts class-level semantic information as a guidance to construct a new semantic guided knowledge graph (SG-KG), which can correct the errors in the existing knowledge graph and accurately model the inter-class relationships. Specifically, our method includes two main steps: noise suppression and semantic enhancement. Noise suppression is used to eliminate noise edges in the knowledge graph, and semantic enhancement is used to connect two classes with strong relations. To promote high efficient information propagation among classes, we develop a novel multi-granularity fusion network (MGFN) that integrates discriminative information from multiple GCN branches. Extensive experiments on the large-scale ImageNet-21K dataset and AWA2 dataset demonstrate that our method consistently surpasses existing methods and achieves a new state-of-the-art result.     

Fine-grained-based multi-feature fusion for occluded person re-identification

Many previous occluded person re-identification(re-ID) methods try to use additional clues (pose estimation or semantic parsing models) to focus on non-occluded regions. However, these methods extremely rely on the performance of additional clues and often capture pedestrian features by designing complex modules. In this work, we propose a simple Fine-Grained Multi-Feature Fusion Network (FGMFN) to extract discriminative features, which is a dual-branch structure consisting of global feature branch and partial feature branch. Firstly, we utilize a chunking strategy to extract multi-granularity features to make the pedestrian information contained in it more comprehensive. Secondly, a spatial transformer network is introduced to localize the pedestrian’s upper body, and then introduce a relation-aware attention module to explore the fine-grained information. Finally, we fuse the features obtained from the two branches to obtain a more robust pedestrian representation. Extensive experiments verify the effectiveness of our method under the occlusion scenario.     

Siamese visual tracking with multilayer feature fusion and corner distance IoU loss

The tracker based on the Siamese network regards tracking tasks as solving a similarity problem between the target template and search area. Using shallow networks and offline training, these trackers perform well in simple scenarios. However, due to the lack of semantic information, they have difficulty meeting the accuracy requirements of the task when faced with complex backgrounds and other challenging scenarios. In response to this problem, we propose a new model, which uses the improved ResNet-22 network to extract deep features with more semantic information. Multilayer feature fusion is used to obtain a high-quality score map to reduce the influence of interference factors in the complex background on the tracker. In addition, we propose a more powerful Corner Distance IoU (intersection over union) loss function so that the algorithm can better regression to the bounding box. In the experiments, the tracker was extensively evaluated on the object tracking benchmark data sets, OTB2013 and OTB2015, and the visual object tracking data sets, VOT2016 and VOT2017, and achieved competitive performance, proving the effectiveness of this method.     

A view and a review of the melting of alkali metal halide crystals

Summary A representational model, proposed to account for the physical changes that accompany the melting of alkali halides, was described in Part 1 [1]. The liquid is portrayed as undergoing continual dynamic structural reorganization of its constituent ions between individual small domains, zones of various regular, crystal-type arrays. These alternative arrangements are stabilized by the enthalpy of melting, which, in liquids, relaxes the restriction for solids that only the single, most stable, crystal structure can be present. The dynamic character of the melt accounts for its fluid character and the loss of long-range order [1, 2]. This model is extended here to consider the phase diagrams of binary, common ion, alkali halide mixtures comprehensively reviewed in [3]. Factors determining whether each of these yields a eutectic, or a solid solution, on cooling are discussed and several trends in the 70-phase diagrams are identified. Eutectic formation, involving maintenance of the liquid state below the melting points of the pure components, is ascribed to the participation, in an extended dynamic equilibrium, of additional domains having the regular structures characteristic of double salts. The known crystalline double binary halides [3], Li/Cs or Rb/F, Cl, Br or I, melt at temperatures well below those of the simpler pure component salts. It is concluded that the set/liq model for melting, proposed in [1, 2], accounts for some important properties of the phase diagrams presented in [3].     

A statistical model for the optimization of dsc performance in the evaluation of drugs for preformulation studies

Differential scanning calorimetry (DSC) is a thermal analytical tool for preformulation studies. Extrapolated melting temperature (T_P) and heat of fusion (ΔH_f) can be used as parameters for optimizing the DSC performance. Two model pharmaceuticals acetaminophen and nicotinamide are used in this study. Using a factorial design for the experimental model and matrix analysis the results, the effect of sample mass, heating rate and the nitrogen flow rate were evaluated on the ΔH_f values and T_P values. Two levels for each of the procedural variables were used as a balanced experimental design with two sample sizes, two heating rates and two nitrogen flow rates. It was found that the change in the heating rate caused significant changes in the ΔH_f values but not the T_p values for acetaminophen. However, no significant effect was found for the T_p value but ΔH_f value was affected to a certain extent for nicotinamide.     

Thermodynamic Study of the Three Isomers of Bromobenzoic Acid

The enthalpies of sublimation and fusion and triple-point temperatures of 2-bromo-. 3-bro-mo- and 4-bromobenzoic acids have been determined precisely by sublimation calorimetry, drop calorimetry and differential thermal analysis. The measurements of sublimation enthalpy of the three acids were made at 333, 348 and 363 K, respectively, using a Tian-Calvet microcalorimeter equipped with Knudsen effusion cells. The derived standard molar enthalpies of sublimation at 298.15 K are (95. 94±0. 41), (99. 20± 0.18), and (103. 08±0. 59) kJ · mol-1for the 2-bromo-, 3-bromo- and 4-bromobenzoic acids, respectively. In addition, the saturated vapour pressure of these compounds was also calculated on the basis of the sublimation experiments. The enthalpy of fusion, the triple-point temperatures and the mole fraction purities of the samples of the investigated substances were measured using the mean temperature version DTA apparatus developed by the CTM of the CNRS in Marseille. The triple-point temperature and the     

熔融反应合成2,2′-联吡啶桥联二（4,4′-联吡啶）类哑铃型化合物

无溶剂反应;联吡啶;固相熔融;哑铃型化合物     

The crystallinity of polyaryl ether ether ketone

Measurement of the degree of crystallinity of the polymer matrix in a composite is complicated by the presence of the reinforcing additive. This is particularly the case in APC-2 in which as much as 70% can be carbon fibre. A First Law procedure, developed for determining the degree of crystallinity of PEEK, which involves direct measurement of the enthalpy changes associated with melting, crystallization and heat capacity changes, has found to be an effective method for the determination of the crystallinity of the PEEK matrix. The procedure has been applied to carbon fibre and glass fibre PEEK composites.