基于Transformer和CNN交错混合的肺结节分割网络

针对肺结节尺寸多样、形状异质化高等问题,提出基于Transformer和卷积神经网络(CNN)交错混合(IMTC)的肺结节分割网络,该网络是一个对称的层次连接网络,具有很强的多尺度特征提取能力。该网络通过集成2种方案分别解决肺结节多尺寸与形状异质化问题:(1)采用感知注意力模块(inception attention module,IAM),通过并联多个不同大小的卷积核来增加浅层网络的感受野组合,以此捕获更为丰富的浅层特征;(2)为获取更具表示能力的高级语义特征,利用由Transformer和CNN组成的基本骨干网络交错提取结节特征,使得全局特征与局部特征充分融合,从而提高结节特征表示的泛化能力和鲁棒性。实验结果表明:本文模型可以准确分割直径较小以及边缘复杂的肺结节,在LUNA16公开数据集上分割性能良好,Dice和IOU分别达到86.15%和76.10%。

V型Transformer的遥感影像障碍物提取方法

遥感影像中的障碍物是地震采集观测系统变观的重要依据之一。传统的人工提取障碍物方法效率低,且易受人为因素影响,难以保证结果的一致性,不适用于复杂地表环境及数量庞大的障碍物。当前通用的卷积神经网络自动提取障碍物方法,由于卷积核的尺寸受限,无法直接进行远距离的语义交互,也不能准确提取具有较大跨度且部分被遮蔽的障碍物(乡间道路、河流等)。为此,提出了基于V型全自注意力网络(MTNet)提取遥感影像障碍物的方法。首先,MTNet采用端到端的V型编码器—解码器结构,通过跳跃连接实现信息交互;其次,用具有远距离建模能力的Mix-Transformer模块取代传统卷积层,提取和重建更准确的障碍物多尺度特征;最后,用轻量的块扩展层取代转置卷积,实现上采样和图像分割,重建障碍物信息。实验结果表明,该网络分割障碍物的精度和速度显著优于现有方法,尤其在道路识别方面,优势更明显。

基于条件对抗增强的Transformer煤矿微震定位方法

随着人工智能技术的发展以及煤矿微震监测系统的广泛应用,越来越多的深度学习模型被应用到煤矿微震事件震源定位问题的求解上。然而,由于目前的微震数据量小且数据单一不足以训练大且深的神经网络模型,而小且浅的神经网络模型也不足以表征受多方因素影响的微震事件的震源,因而导致了定位模型定位精度低和鲁棒性弱,在实际生产生活中表现较差,严重地阻碍了深度学习模型在微震定位领域上的发展。针对上述问题,提出一种基于条件对抗增强的Transformer煤矿微震定位方法CGAN-Transformer,该方法首先通过一个CGAN架构的网络模型将数据量少且单一的微震数据增强成数据量庞大且具有一定多样性的微震数据;其次,利用Transformer编码器层将微震波形数据转换为特征数据后再利用其注意力机制进一步学习微震波形数据深层次特征和复杂的站间依赖关系,同时也利用高斯分布随机变量抵消了不同地质条件对定位精度的影响;最后,通过引入混合密度输出层获取高斯分布参数,计算最优的震源位置。在智利和辽宁某矿数据集上的实验结果验证了该方法的有效性,结果表明该方法所获得的震中误差与震源误差均优于其他方法,在两个数据集上的定位误差分别降低了38%和12%,达到了提高震源定位精度和定位模型鲁棒性的目的。

ARoad Segmentation Model Based on Mixture of the Convolutional Neural Network and the Transformer Network

Convolutional neural networks(CNN)based on U-shaped structures and skip connections play a pivotal role in various image segmentation tasks.Recently,Transformer starts to lead new trends in the image segmentation task.Transformer layer can construct the relationship between all pixels,and the two parties can complement each other well.On the basis of these characteristics,we try to combine Transformer pipeline and convolutional neural network pipeline to gain the advantages of both.The image is put into the U-shaped encoder-decoder architecture based on empirical combination of self-attention and convolution,in which skip connections are utilized for localglobal semantic feature learning.At the same time,the image is also put into the convolutional neural network architecture.The final segmentation result will be formed by Mix block which combines both.The mixture model of the convolutional neural network and the Transformer network for road segmentation(MCTNet)can achieve effective segmentation results on KITTI dataset and Unstructured Road Scene(URS)dataset built by ourselves.Codes,self-built datasets and trainable models will be available on https://github.com/xflxfl1992/MCTNet.

Some properties of mixed radial mean bodies on integral transform

As a generalization of RpK, the mixed radial p-th mean bodies Rp（K,Lμ） have many properties similar to RpK. In this paper, some properties of Rp（K, L,μ） such as monotonicity, transform properties under volume-preserving linear transformations, etc. were studied. By using the polar coordinate formula for volume and Fubini＇s theorem, V（Rn（K, L,μ）） = V（L） was obtained.

Research on the Digital Transformation of English Teaching in Vocational Education

With the advancement of technology,exploring the impact of digital transformation on vocational education English teaching has become crucial.This study aims to investigate the effectiveness of digital transformation in English teaching in vocational education in China by exploring students’and teachers’attitudes,views,and experiences on the use of digital technology in English teaching.This study employed a mixed method of qualitative and quantitative analysis.The research results indicate that digital transformation has had a positive impact on vocational education English teaching,as it enhances the teaching process,promotes communication and collaboration,and increases students’enthusiasm and participation.However,implementing digital transformation in vocational education English teaching also poses challenges,including a lack of resources,infrastructure,and training.This study provides an in-depth understanding of the advantages and challenges of digital transformation in vocational education English teaching and proposes strategies to improve the implementation of digital technology in this context.

基于混合信号多域特征和Transformer的干扰识别

针对无线通信信道易受到蓄意射频信号干扰问题,提出了一种从混合信号中识别干扰类型的方法。通过改进经典Transformer结构,形成新型网络模型Multidomain-former,以提取多域特征和识别信号干扰类型。首先,通过特定的序列划分机制对输入频谱进行预处理,并通过线性嵌入和位置编码保留原始顺序特征;其次,设计了逆傅氏变换和傅氏变换结合的编码模块,使Multidomain-former能同时提取频域和时域特征。使用通用仪器和收发天线搭建了无线收发信道,在不同干信比条件下对混合信号频谱进行采集,得到训练集和测试集。干扰对比实验通过所提Multidomain-former网络模型完成,并将经典的Transformer结构和其他常见的深度学习模型与所提网络模型进行了对比。对比实验结果表明,在干信比小于10 dB时,所提模型性能相较于经典Transformer在识别正确率方面有2%~3%的提升;在干信比等于-5 dB时,所提模型以最少参数量和次低计算复杂度获得了比另外5种基准网络高3.0%~9.3%的识别率。

Estimation of Wave Crest Amplitudes Distribution and Freak Wave Occurrence in A Short Crested Mixed Sea

In this study we have for the first time proposed a novel transformed linear simulation method for the estimation of wave crest amplitudes distribution and freak wave occurrence in a short crested mixed sea with a bimodal 3D spectrum. For implementing the proposed transformed linear simulation method, a Hermite transformation model expressed in a monotonic cubic polynomial has been constructed so that the first four moments of the original true process match the corresponding moments of the transformed model. The proposed novel simulation method has been applied to forecast the freak wave occurrence in two short crested mixed sea states, one with a directional wave spectrum based on the measured surface elevation data at the coast of Yura, and the other one with a typical directional bimodal Torsethaugen wave spectrum. It is shown in the two cases that the proposed novel simulation method can offer more accurate forecasting results than those obtained from the traditional linear simulation method or by using Rayleigh distribution model. It is also demonstrated in this article that the proposed novel simulation method is more efficient than the nonlinear simulation method.

Darcy-Forchheimer Hybrid Nano Fluid Flow with Mixed Convection Past an Inclined Cylinder

This article aims to investigate the Darcy Forchhemier mixed convection flow of the hybrid nanofluid through an inclined extending cylinder.Two different nanoparticles such as carbon nanotubes(CNTs)and iron oxide Fe3O4 have been added to the base fluid in order to prepare a hybrid nanofluid.Nonlinear partial differential equations for momentum,energy and convective diffusion have been changed into dimensionless ordinary differential equations after using Von Karman approach.Homotopy analysis method(HAM),a powerful analytical approach has been used to find the solution to the given problem.The effects of the physical constraints on velocity,concentration and temperature profile have been drawn as well for discussion purpose.The numerical outcomes have been carried out for the drag force,heat transfer rate and diffusion rate etc.The Biot number of heat and mass transfer affects the fluid temperature whereas the Forchhemier parameter and the inclination angle decrease the velocity of the fluid flow.The results show that hybrid nanofluid is the best source of enhancing heat transfer and can be used for cooling purposes as well.

Numerical modeling of 3D DC resistivity method in the mixed space-wavenumber domain

Forward modeling is the basis of inversion imaging and quantitative interpretation for DC resistivity exploration.Currently,a numerical model of the DC resistivity method must be finely divided to obtain a highly accurate solution under complex conditions,resulting in a long calculation time and large storage.Therefore,we propose a 3D numerical simulation method in a mixed space-wavenumber domain to overcome this challenge.The partial differential equation about abnormal potential is transformed into many independent ordinary differential equations with different wavenumbers using a 2D Fourier transform along the x axis and y axis direction.In this way,a large-scale 3D numerical simulation problem is decomposed into several 1D numerical simulation problems,which significantly reduces the computational and storage requirements.In addition,these ordinary 1D differential equations with different wavenumbers are independent of each other and high parallelelism of the algorithm.They are solved using a finite-element algorithm combined with a chasing method,and the obtained solution is modified using a contraction operator.In this method,the vertical direction is reserved as the spatial domain,then grid size can be determined flexibly based on the underground current density distribution,which considers the solution accuracy and calculation efficiency.In addition,for the first time,we use the contraction operator in the integral equation method to iterate the algorithm.The algorithm takes advantage of the high efficiency of the standard Fourier transform and chasing method,as well as the fast convergence of the contraction operator.We verified the accuracy of the algorithm and the convergence of the contraction operator.Compared with a volume integral method and goal-oriented adaptive finite-element method,the proposed algorithm has lower memory requirements and high computational efficiency,making it suitable for calculating a model with large-scale nodes.Moreover,different examples are used to verify the high adaptability and parallelism of the proposed algorithm.The findings show that the 3D numerical simulation method of DC resistivity method in a mixed space-wavenumber domain is highly efficient,precise,and parallel.

A new method for the estimation of oceanic mixed-layer depth using shipboard X-band radar images

Shipboard X-band radar images acquired on 24 June 2009 are used to study nonlinear internal wave characteristics in the northeastern South China Sea.The studied images show three nonlinear internal waves in a packet.A method based on the Radon Transform technique is introduced to calculate internal wave parameters such as the direction of propagation and internal wave velocity from backscatter images.Assuming that the ocean is a two-layer finite depth system,we can derive the mixed-layer depth by applying the internal wave velocity to the mixed-layer depth formula.Results show reasonably good agreement with in-situ thermistor chain and conductivity-temperature-depth data sets.

A NONLINEAR LAVRENTIEV-BITSADZE MIXED TYPE EQUATION

In this paper the Tricomi problem for a nonlinear mixed type equation is studied. The coefficients of the mixed type equation are discontinuous on the line, where the equation changes its type. The existence of solution to this problem is proved. The method developed in this paper can be applied to study more difficult problems for nonlinear mixed type equations arising in gas dynamics.

Inequalities for L_p Mixed Intersection Bodies

Haberl and Ludwig defined the notions of L_p intersection bodies. In this paper,we introduce the L_p mixed intersection bodies, and establish some geometric inequalities for L_p mixed intersection bodies. Furthermore, the Busemann-Petty type problem for L_p mixed intersection bodies are shown.

A Mixed-control Mechanism Model of Proeutectoid Ferrite Growth under Non-equilibrium Interface Condition in Fe-C Alloys

By combining the α/γ interface migration and the carbon diffusion at the interface in Fe-C alloys, a mathematical model is constructed to describe the mixed-control mechanism for proeutectoid ferrite formation from austenite. In this model, the α/γ interface is treated as non-equilibrium interface, i.e., the carbon concentration of austenite at γ/α interface is obtained through theoretical calculation, instead of that assumed as the local equilibrium concentration. For isothermal precipitation of ferrite in Fe-C alloys, the calculated results show that the rate of interface migration decreases monotonically during the whole process, while the rate of carbon diffusion from γ/α interface into austenite increases to a peak value and then decreases. The process of ferrite growth may be considered as composed of three stages: the period of rapid growth, slow growth and finishing stage. The results also show that the carbon concentration of austenite at γ/α interface could not reach the thermodynamic equilibrium value even at the last stage of ferrite growth.

Digital Image Watermarking Based on Mixed Error Correcting Code

In this paper, we present a novel technique based on a mixed Error Correcting Code(ECC)-the convolutional code and the repetition code to enhance the robustness of the embedded watermark. Before embedding, the binary watermark is scanned to one-dimension sequence and later inputted into the (3, 1, 2) convolutional encoder and (3, 1) repetition encoder frame by frame, which will improve the error correcting capability of decoder. The output code sequence is scanned to some matrixes as the new watermark messages. The watermarking is selected in low frequency band of the Discrete Wavelet Transform (DWT) and therefore it can resist the destruction of image processing. Experimental results are presented to demonstrate that the robustness of a watermark with mixed ECC is much higher than the traditional one just with repetition coding while suffering JPEG lossy compression, salt and pepper noise and center cutting processing.

A novel image fusion algorithm based on 2D scale-mixing complex wavelet transform and Bayesian MAP estimation for multimodal medical images

In this paper,we propose a new image fusion algorithm based on two-dimensional Scale-Mixing Complex Wavelet Transform(2D-SMCWT).The fusion of the detail 2D-SMCWT cofficients is performed via a Bayesian Maximum a Posteriori(MAP)approach by considering a trivariate statistical model for the local neighboring of 2D-SMCWT coefficients.For the approx imation coefficients,a new fusion rule based on the Principal Component Analysis(PCA)is applied.We conduct several experiments using three different groups of multimodal medical images to evaluate the performance of the proposed method.The obt ained results prove the superiority of the proposed method over the state of the art fusion methods in terms of visual quality and several commonly used metrics.Robustness of the proposed method is further tested against different types of noise.The plots of fusion met rics establish the accuracy of the proposed fusion method.

A Multiphase Mixed Methods Approach to the Internationalization of Higher Education in South Africa:Outline of a Discursive Research Framework

Higher education internationalization(HEI)has become an important priority for global decision makers and education providers.Persuaded by globalization,HEI is a political strategy and economic prospect for countries,governments,and universities to position themselves in global markets.The purpose of this paper is to provide a methodological pathway to develop a dynamic,reformed South African HEI framework based on the transformative drive(i.e.,adjustment of change,learning,shared knowledge,internationalization,globalization an institutionalized memory)of HEI in China.The interpretivist paradigm constitutes the basic research approach.Quantitative and qualitative data will be collected,analyzed,and integrated using a multiphase mixed methods approach.This approach consists of three phases including content analysis(Phase I),in-depth interviews(Phase II),and surveys(Phase III),which answers the focal research question.The significance of this research includes creating a better understanding of South African higher education(HE)policy issues from a transformational perspective,contributing to theory on how internationalization affects HE’s position in economic and social development and broadening knowledge and understanding of the unique circumstances and challenges of HEI in developing countries,particularly in countries seeking to position themselves in the globalized knowledge economy.

Study of the Stacking Sequences of an Irregular Mixed-Layer Illite/Smectite (I/S) Clay Mineral—with a Discussion on the Existence of Minerals with Two-Dimentional Lattice and One-Dimentional Quasi-Lattice

There are two aspects in the study of irregular mixed-layer clay minerals: one is the kinds and ratios of their basic structural unit layers and the other is the junction probabilities of the unit layers. Irregular mixed-layer illite/smectite clay minerals (I/S) are widespread in nature. While studying the clay minerals from the Permian-Triassic (P/T) boundary bed, the authors found that I/S clay minerals are developed in all P/T boundary clay layers in areas from the northwest to southeast of China. Systematic mineralogical studies of the I/S clay minerals from Hunan, Hubei, Sichuan and Zhejiang were made by means of X-ray, infrared spectroscopic, electron microscopic and chemical analyses and a deepened study of the stacking sequences of their structural unit layers was conducted by the MacEwan one—dimentional direct Fourier transform. It was found that the stacking of the illite and smectite crystal layers along the c axis can be derived from Fibonacci sequences. Hence, the authors propose that such I/S clay minerals are possessed of two—dimentional crystal lattice and one—dimentional quasicrystal lattice.

数字化转型、服务化融合与制造业企业绩效提升——兼论经济政策不确定性的调节效应

引入经济政策不确定性作为调节变量,从嵌入式服务化和混入式服务化的双重融合视角,研究数字化转型提升制造业企业绩效的理论机制和作用路径。在此基础上,采用2008—2020年制造业A股上市公司数据对理论机制进行实证检验。研究结果显示:数字化转型能够提升企业绩效,嵌入式服务化发挥了正向中介作用,但在区域分布、所有制结构和技术水平等方面具有明显的异质性;数字化转型能够抑制混入式服务化中制造业企业的“脱实向虚”行为;经济政策的不确定性会增强数字化转型对嵌入式服务化的推动作用和对混入式服务化的抑制作用。

非国有资本参股对国有企业数字化转型的影响研究

国有企业推进数字化转型在促进自身发展升级时也能产生广泛的社会价值。基于2007—2020年上市企业数据,对非国有资本参股与国有企业数字化转型的关系及其内在机理进行实证分析。研究结果表明:国有企业在数字化转型中面临情境式困境,非国有资本参股通过加强企业创新动力、激发高管企业家精神与完善人才激励机制促进数字化转型,从而实现国有经济的高质量发展,但其效果会因国有企业特征、非国有资本特征以及外部环境特征有所差异。