Clothing Parsing Based on Multi-Scale Fusion and Improved Self-Attention Mechanism

Due to the lack of long-range association and spatial location information,fine details and accurate boundaries of complex clothing images cannot always be obtained by using the existing deep learning-based methods.This paper presents a convolutional structure with multi-scale fusion to optimize the step of clothing feature extraction and a self-attention module to capture long-range association information.The structure enables the self-attention mechanism to directly participate in the process of information exchange through the down-scaling projection operation of the multi-scale framework.In addition,the improved self-attention module introduces the extraction of 2-dimensional relative position information to make up for its lack of ability to extract spatial position features from clothing images.The experimental results based on the colorful fashion parsing dataset(CFPD)show that the proposed network structure achieves 53.68%mean intersection over union(mIoU)and has better performance on the clothing parsing task.

Hierarchical multihead self-attention for time-series-based fault diagnosis

Fault diagnosis is important for maintaining the safety and effectiveness of chemical process.Considering the multivariate,nonlinear,and dynamic characteristic of chemical process,many time-series-based data-driven fault diagnosis methods have been developed in recent years.However,the existing methods have the problem of long-term dependency and are difficult to train due to the sequential way of training.To overcome these problems,a novel fault diagnosis method based on time-series and the hierarchical multihead self-attention(HMSAN)is proposed for chemical process.First,a sliding window strategy is adopted to construct the normalized time-series dataset.Second,the HMSAN is developed to extract the time-relevant features from the time-series process data.It improves the basic self-attention model in both width and depth.With the multihead structure,the HMSAN can pay attention to different aspects of the complicated chemical process and obtain the global dynamic features.However,the multiple heads in parallel lead to redundant information,which cannot improve the diagnosis performance.With the hierarchical structure,the redundant information is reduced and the deep local time-related features are further extracted.Besides,a novel many-to-one training strategy is introduced for HMSAN to simplify the training procedure and capture the long-term dependency.Finally,the effectiveness of the proposed method is demonstrated by two chemical cases.The experimental results show that the proposed method achieves a great performance on time-series industrial data and outperforms the state-of-the-art approaches.

Polarization characteristics and controllability mechanism of passive scattering elements

Polarization feature is one of the important features of radar targets,which has been used in many fields.In this paper,the grid models of some typical foreign moving targets are constructed on the simulation platform,such as glider,cruiser,fixed wing aircraft,and rotorcraft.The electromagnetic scattering characteristics of the moving platforms under the incidence of circular polarization waves are calculated.The typical polarization characteristics which the orthogonal and in-phase components have in the echoes are analyzed and proved.Based on the polarization scattering matrix(PSM)theory,from the point of view of the physical reproduction,the technical status quo that the existing technical approaches are difficult to realize the passive simulation of polarization characteristic of the target is summarized.To solve this problem,combined with the vector synthesis law,the realization mechanism of controllable polarization characteristic of target echoes is proposed,the analytical expressions of polarization control matrix and polarization ratio are deduced,and the controllability of polarization ratio feature in the case of circular polarization is verified by simulation calculation.

Keyphrase Generation Based on Self-Attention Mechanism

Keyphrase greatly provides summarized and valuable information.This information can help us not only understand text semantics,but also organize and retrieve text content effectively.The task of automatically generating it has received considerable attention in recent decades.From the previous studies,we can see many workable solutions for obtaining keyphrases.One method is to divide the content to be summarized into multiple blocks of text,then we rank and select the most important content.The disadvantage of this method is that it cannot identify keyphrase that does not include in the text,let alone get the real semantic meaning hidden in the text.Another approach uses recurrent neural networks to generate keyphrases from the semantic aspects of the text,but the inherently sequential nature precludes parallelization within training examples,and distances have limitations on context dependencies.Previous works have demonstrated the benefits of the self-attention mechanism,which can learn global text dependency features and can be parallelized.Inspired by the above observation,we propose a keyphrase generation model,which is based entirely on the self-attention mechanism.It is an encoder-decoder model that can make up the above disadvantage effectively.In addition,we also consider the semantic similarity between keyphrases,and add semantic similarity processing module into the model.This proposed model,which is demonstrated by empirical analysis on five datasets,can achieve competitive performance compared to baseline methods.

MOLECULAR POLARIZATION INDUCTION IN CONCERTED MECHANISM OF ALKOXY MIGRATION

The rearrangement of α-substituted phenyl-α, α′-dimethoxypropanones (R1) was recently proved to he an [1, 3] siqmatropic migration of methoxy group by study of LFER, solvent effect, crossover experiment, and by other means. The rearrangement catalysed by trace of totuenesulfonic acid gave a negligible value of the Hammett reaction constant. Further experimental results show that the electronic effect of substituent on the aromatic ring for the rearrangement in neutral medium is much greater (Table 1). We propose sigmatropic migration

An Equivalent Substitute Strategy for Constructing 3D Ordered Porous Carbon Foams and Their Electromagnetic Attenuation Mechanism

Three-dimensional(3D)ordered porous carbon is generally believed to be a promising electromagnetic wave(EMW)absorbing material.However,most research works targeted performance improvement of 3D ordered porous carbon,and the specific attenuation mechanism is still ambiguous.Therefore,in this work,a novel ultra-light egg-derived porous carbon foam(EDCF)structure has been successfully constructed by a simple carbonization combined with the silica microsphere template-etching process.Based on an equivalent substitute strategy,the influence of pore volume and specific surface area on the electromagnetic parameters and EMW absorption properties of the EDCF products was confirmed respectively by adjusting the addition content and diameter of silica microspheres.As a primary attenuation mode,the dielectric loss originates from the comprehensive effect of conduction loss and polarization loss in S-band and C band,and the value is dominated by polarization loss in X band and Ku band,which is obviously greater than that of conduction loss.Furthermore,in all samples,the largest effective absorption bandwidth of EDCF-3 is 7.12 GHz under the thickness of 2.13 mm with the filling content of approximately 5 wt%,covering the whole Ku band.Meanwhile,the EDCF-7 sample with optimized pore volume and specific surface area achieves minimum reflection loss(RL_(min))of−58.08 dB at 16.86 GHz while the thickness is 1.27 mm.The outstanding research results not only provide a novel insight into enhancement of EMW absorption properties but also clarify the dominant dissipation mechanism for the porous carbon-based absorber from the perspective of objective experiments.

Polarization properties in a prism laser gyro with mechanical dither bias

Theoretical analysis and experimental research on the polarization properties of output light in gyro are carried out to investigate the phenomenon that the amplitude of an output voltage signal is modulated by dither bias in laser gyros consisting of totally reflecting prisms. Taking the effect of prism stress birefringence into account, an analytical formula of the output light intensity in the gyro and the relationship between the polarization parameter and the amplitude modulation of the output signal are obtained and discussed. For the first time, the polarized power value of the output light is adopted as a basis to estimate the output signal amplitude fading extent of laser gyros. Experimental results demonstrate that when the value of polarized power of output light is below 25.5% of that in ideal static case, the standard error is over 0.0337 dBm, and the displacement extent of the prism is higher than 53% of the radius of the beam waist in the gyro cavity, the amplitude modulation extent of gyro output signal can reach up to 16%, which badly influences the measurement accuracy of the laser gyro. Using this polarized power detecting measurement method can repair the gyro immediately during its fabrication process, improve the testing and production efficiency and shorten the product development cycle.

Microstructures, mechanical properties, corrosion, and biocompatibility of extruded Mg-Zr-Sr-Ho alloys for biodegradable implant applications

In this study,the microstructures,mechanical properties,corrosion behaviors,and biocompatibility of extruded magnesium-zirconiumstrontium-holmium(Mg-Zr-Sr-Ho)alloys were comprehensively investigated.The effect of different concentrations of Ho on the microstructural characteristics,tensile and compressive properties,corrosion resistance,and biocompatibility were investigated.The microstructures of the extruded Mg-1Zr-0.5Sr-xHo(x=0.5,1.5,and 4 wt.%)alloys consisted ofα-Mg matrix,fineα-Zr particles,and intermetallic phase particles of Mg_(17)Sr_(2) and Ho_(2)Mg mainly distributed at the grain boundaries.Extensive{1012}tensile twins were observed in the partially recrystallized samples of Mg-1Zr-0.5Sr-0.5Ho and Mg-1Zr-0.5Sr-1.5Ho.Further addition of Ho to 4 wt.%resulted in a complete recrystallization due to activation of the particle stimulated nucleation around the Mg_(17)Sr_(2) particles.The evolution of a rare earth(RE)texture was observed with the Ho addition,which resulted in the weakened basal and prismatic textures.Furthermore,a drastic increase of 200%in tensile elongation and 89%in compressive strain was observed with Ho addition increased from 0.5 to 4 wt%,respectively.The tension-compression yield asymmetry was significantly decreased from 0.62 for Mg-1Zr-0.5Sr-0.5Ho to 0.98 for Mg-1Zr-0.5Sr-4Ho due to the weakening of textures.Corrosion analysis of the extruded Mg-Zr-Sr-Ho alloys revealed the presence of pitting corrosion.A minimum corrosion rate of 4.98 mm y^(−1) was observed in Mg-1Zr-0.5Sr-0.5Ho alloy.The enhanced corrosion resistance is observed due to the presence of Ho_(2)O_(3) in the surface film which reduced galvanic effect.The formation of a stabilized surface film due to the Ho_(2)O_(3) was confirmed through the electrical impedance spectroscopy and XPS analysis.An in vitro cytotoxicity assessment revealed good biocompatibility and cell adhesion in relation to SaOS2 cells.

Mechanical properties of welded joints in vertical-up variable polarity plasma arc welding of 2219 aluminum alloy

The mechanical properties of the 2219 aluminum alloy welded joints by vertical-up variant polarity plasma arc welding (VPPAW) with keyhole were analyzed. The tensile strength of welded joints can reach 55% of that of base metal. The microstructure of weld joints was studied to explain the mechanical properties of the weld. The results show that the intensity of weld center is less than that of HAZ, and the intensity of weld center is the worst. So an advanced welding procedures is presented to reduce the heat input and increase the welding speed. The results in this procedures show that the tensile strength of welded joints reaches more than 60% of that of the parent metal, and the specific elongation does not descend.

Mechanical Force Drives the Polarization and Orientation of Collective Cells

Collective cell groups are organized to form specific patterns that play an important role in various physiological and pathological processes,such as tissue morphogenesis,wound healing,and cancer invasion.Compared to the behaviors of single cells that have been studied intensively from many aspects(cell migration,adhesion,polarization,proliferation,etc.)and at various length scales(molecular,subcellular,and cellular),the behaviors of multiple cells are less well understood,particularly from a quantitative perspective.In this talk,we present our recent studies of collective polarization and orientation of multiple cells through both experimental measurement and theoretical modeling,including cell behavior on/in 2D and 3D substrate/tissue.We find that collective cell behavior,including polarization,alignment,and migration,is closely related to local stress states in cell layers or tissue,which demonstrates the crucial role of mechanical forces in living organisms.Specifically,cells demonstrate preferential polarization and alignment along the maximum principal stress in the cell layer,and the cell aspect ratio increases with in-plane maximum shear stress,suggesting that the maximum shear stress is the underlying driving force of cell polarization and orientation.This theory of stress-driven cell behaviors of polarization and orientation provides a new perspective for understanding cell behaviors in living organisms and a guideline for tissue engineering in potential biomedical applications.Strikingly,we note that with regard to the polarization and alignment of collective cells,a typical feature of cell morphology is that the cells generally align along the edge of the pattern,which was called edge effect or boundary effect by assuming that the edge plays a role in cell alignment due to a phenomenon of chemistry.However,the edge effect is an obscure explanation.Here we showed that the edge effect could be explained by the theory of stress-driven cell behavior,i.e.,inplane stress-driven cell polarization and alignment.That is,the cell layer has a stress-free boundary condition at the edge,and thus the direction of the maximum principal stress should be precisely along the edge.According to the theory of stress-driven cell polarity,the cells then preferentially align with the edge of the cell layer,independently of the geometry of the pattern.Once there is a force-free condition at the edge or the boundary,the cells align along the edge of the pattern.Otherwise,the cell may not align with the edge;for example,the cells preferentially align in the radial direction of the wound because of the presence of the contractile force by the actin ring at the wound edge,which is in contradiction with the so-called edge effect but consistent with our theory of stress-driven cell polarity.

Quantitative Study of Lunisolar Precession Mechanism

According to Xiaodong Song et al.’s study about difference rotation in the Earth’s inner core, a physical model of lunar motion changing the angular momentum of the Earth’s circles is presented in this work. The Lunar motion makes there be a huge gap existing in the core angular momentum between far lunar hemisphere and near lunar hemisphere. The gap results in the difference rota-tion in the Earth’s inner core, meanwhile, it increases the angular velocity of far lunar hemisphere and decreases the angular velocity of near lunar hemisphere. The Earth’s liquid outer core gene-rates an astro-geodynamical effect including the difference rotation among the Earth’s circles. It is found that when the moon moves into the apogee or the perigee of the lunar orbit, and the moon phase is the upper and lower chords (i.e. semi-diameter place), the true anomaly of the moon will change from 270-degree back (or forward) to 90-degree;this results in the mantle shell of the Earth westward, and forms lunisolar precession, the vernal equinox westward and Chandler polar motion.

基于双重注意力机制生成对抗网络的偏振图像融合

针对单一强度图像缺少偏振信息,在恶劣天气条件下无法提供充足场景信息的问题,本文提出了一种基于双重注意力机制生成对抗网络用于强度图像和偏振度图像进行融合。算法网络由一个包含编码器、融合模块和解码器的生成器和一个鉴别器组成。首先源图像输入到生成器的编码器中,经过一个卷积层和密集块进行特征提取,然后通过含有注意力机制的纹理增强融合模块中进行特征融合,最后通过解码器得到融合图像。鉴别器主要由两个卷积模块和两个注意力模块组成,在网络训练过程中,通过不断博弈,迭代优化生成器网络参数,使生成器输出既保留偏振度图像的稀疏特征又不损失强度图像信息的高质量融合图像。实验表明,该方法得到的融合图像在主观上纹理信息更丰富,更符合人眼的视觉感受,并且在客观评价指标中SD提升约18.5%,VIF提升约22.4%。

复杂管道浓差腐蚀机制深入分析

在关于管道的流动加速腐蚀研究中,浓差腐蚀这一机制一直被忽视,这是因为这种腐蚀的出现需要一定的条件,即流体中的氧的分布存在较大的差异。为了揭示这种机制的腐蚀过程,本文以复杂形状管道为例,采用数值模拟技术建立了关于浓差腐蚀的数学模型。将管道壁面附近的流体粘性底层作为研究对象,对其进行离散化处理,然后根据基尔霍夫第二定律建立了关于极化后的腐蚀电流的离散方程组,并对其进行求解,得到了浓差腐蚀条件下,离子导电层的电流分布。结果表明:浓差腐蚀会引起腐蚀电流的极大的改变,使管路各部位的腐蚀情况有所改变,如果不考虑浓差腐蚀,对管路的腐蚀情况会产生错误判断。

基于液晶偏振光栅的大口径激光光束扫描系统

非机械式激光光束扫描系统具有体积小、重量轻、易集成等优势,在激光雷达、激光武器和激光通信等领域具有较大的应用前景,其通光口径决定着设备的工作距离、扫描精度和目标摧毁能力。基于液晶偏振光栅设计了一种全电控制的大口径激光光束扫描系统。采用液晶偏振光栅和液晶空间光调制器的协同工作方式,实现光束的大角度连续扫描;设计了光束扫描系统,并重点分析了级联液晶偏振光栅组的控制策略,给出了具体控制方法。最后,进行了激光光束的初步扫描实验,实现了70 mm口径光束在−16.575°~16.575°的偏转。该研究结果为激光光束的非机械式扫描控制提供了技术支撑,并在大口径激光扫描系统应用方面进行了初步探索。

中药调控小胶质细胞极化改善抑郁症研究进展

抑郁症是危害人类健康生活的常见精神类疾病,其发病机制尚未明确。目前认为神经炎症假说与抑郁症病理生理关系密切。该假说提出,在抑郁症病程中,持续的刺激过度激活小胶质细胞(Microglia, MG)的M1极化表型,通过分泌大量促炎因子,造成脑部病理损伤,加重抑郁的病理发展。因而通过调控MG极化表型,可能是改善抑郁症的一种有效干预策略。临床会应用抗炎药物改善患者抑郁情绪,虽然有一定的疗效,但存在适用范围窄、不良反应明显等局限性。中药具有多靶点、多途径的特点,其药性内容丰富且复杂。近年来,研究发现诸多中药通过修复线粒体功能障碍、促进脑源性神经营养因子(Brain-derived neurotrophic factor, BDNF)、调节蛋白表达、调节神经炎症反应等机制发挥了抗抑郁作用。因此,以MG为治疗靶点,对中药和复方减轻神经炎症改善抑郁症的文献研究进行梳理和探讨,期望为临床治疗抑郁症提供新的治疗思路。

一种游动傅科机械编排的捷联惯性间接极区导航算法

针对指北方位、游移方位等机械编排的惯性导航系统在极区附近出现计算溢出、导航误差发散、方向基准缺失等问题,本文提出了一种基于游动傅科机械力学编排的捷联惯性间接极区导航算法。首先,推导了传统坐标系和横向坐标系下的游动傅科机械编排位置、速度和姿态转换关系,并设计了完整的捷联惯性极区导航流程。其次,在详细分析载体沿经线穿越极点时的导航解算性能基础上,提出了一种游动傅科机械编排的捷联惯性间接极区导航算法。实验结果表明,该算法能够实现传统导航与横向导航方式的平滑切换,并借助横向导航信息解决载体穿越极点时无法导航的难题。

铝合金VPPAW穿孔焊接匙孔闭合处的微观组织与力学性能

铝合金因其密度低,比强度高等特性,被广泛应用到航空航天、军事工业等领域.与其他焊接工艺相比,变极性等离子弧焊接在大型铝合金结构件加工方面具有突出优势.针对轻金属合金等离子弧焊接过程中匙孔难以自由闭合的问题,提出了气电分离时序脉冲等离子弧环焊缝收弧方法,对比研究了环焊缝收弧方法与手工钨极惰性气体保护焊(tungsten inert gas welding, TIG焊)填补法焊缝匙孔闭合处的微观组织与力学性能.结果表明,通过调节焊接起弧、收弧策略,实现了等离子弧匙孔自由闭合,焊缝收弧区成形良好,组织为均匀的等轴晶;抗拉强度为329.9 MPa,断后伸长率为16.3%,比TIG填补法的抗拉强度提高了22.91%,断后伸长率提高了55.24%.平均硬度值为78.9 HV0.2,热影响区硬度呈下降趋势,熔合线处硬度呈升高趋势,焊缝区硬度波动范围为75~81 HV0.2.

联合高阶目标感知与相似匹配的目标跟踪算法

视觉目标跟踪算法利用自注意力机制增强上下文联系,但面对复杂场景时,自注意力机制中的相关性易发生失配,为此提出一种联合高阶目标感知与相似匹配的目标跟踪算法。构建高阶目标感知模型,针对自注意力机制中的一阶自注意图,利用坍塌的极化过滤方式进行空间和通道维度的正交化建模,优化内部相关性;同时组合非线性拟合函数避免坍塌引起的信息损失,进而获得高阶自注意图,捕获具有高阶上下文信息的感知特征。通过不同维度分解目标的感知特征来细化匹配区域,抑制背景噪声并约束当前帧的响应图,提高网络的判别力。在OTB100和UAV123基准的实验结果表明,所提算法有更好的跟踪性能,可以有效应对相似干扰等问题。

北虫草多糖诱导巨噬细胞极化的机制研究

为了检测北虫草多糖对巨噬细胞极化状态的影响,研究采用外源北虫草多糖刺激小鼠巨噬细胞RAW264.7后,检测巨噬细胞极化状态的变化,探究北虫草多糖诱导巨噬细胞极化的分子机制。结果表明,北虫草多糖刺激可以促进培养巨噬细胞的M1型极化,且极化过程活化了参与炎症免疫应答的NF-κB通路,这一结果初步揭示了北虫草多糖诱导巨噬细胞极化的机制,为进一步深入理解北虫草多糖调控机体免疫力提供理论依据。