金融机构ESG投资的“漂绿”与“反漂绿”——基于DWSGroup的案例分析

深化绿色金融发展、助力经济绿色发展,是实现“双碳”目标的重要途径,特别是在ESG理念盛行的今天,金融机构的ESG投资行为因其深远影响而颇受投资者青睐,但层出不穷的“漂绿”现象也牢牢制约着绿色金融的发展,制约着ESG投资的发展。西方国家作为绿色金融发展先驱,从信息披露、制度监管、实践探索等方面对“漂绿”现象进行了研究和治理。本文在学习其先进经验的基础上,基于舞弊三角理论对DWSGroup案例进行分析,从压力、机会和自我合理化因素角度对金融机构ESG投资“漂绿”行为的动因进行分析,并据此提出我国“反漂绿”体系框架,以期为绿色金融的平稳发展和“双碳”目标的实现作出贡献。

基于改进VGG16的自编码器视频异常检测算法

在使用自编码器结构的神经网络处理视频异常检测任务时,U-Net风格的自编码器由于编码器层数深度过浅,导致在面对复杂的数据集时,不能充分抽取更多有用的特征信息。同时,在训练模型时使用MSE(均方误差),仅考虑了预测帧与真实帧之间的像素级相似性,对于复杂场景,像素级相似性可能无法准确判断预测帧与真实帧之间的相似性。针对以上问题,对基于U-Net风格的自编码器进行改进,提出了一种使用改进的VGG16作为编码器的视频异常检测算法,同时在均方误差的基础上添加结构相似性(SSIM)损失函数。改进的VGG16去掉了全连接层,并加入了残差连接防止特征退化,添加SSIM在计算像素级相似性的同时计算图像的亮度、对比度和结构等方面的相似性来优化网络。实验结果表明,改进后的算法,在Ped2数据集上检测效果达到95.91%,在Avenue数据集上检测效果达到84.89%,与改进前的方法相比分别提高了0.80%和0.19%,验证了所提方法的有效性。

基于VGG-UNet的食用菌菌丝体表型参数自动测量方法

食用菌菌丝体表型特征是食用菌种质资源评价和科学育种的重要依据。针对传统阈值分割方法提取菌丝体区域易受到光照不均、菌丝体不规则生长和培养皿内产生代谢物等因素干扰的问题,制作食用菌菌丝体图像数据集,并提出一种基于深度学习的食用菌菌丝体表型参数自动测量方法。将U-Net网络编码器部分替换为VGG16的前13个卷积层,引入预训练权重,构建适用于菌丝体分割的VGG-UNet模型。测试集上对比实验表明,该模型的平均交并比达到98.18%,比原始U-Net模型高0.93个百分点。经该模型获取菌丝体分割图像后,利用OpenCV相关函数计算菌丝体的半径、周长、面积、覆盖度、圆整度这5个表型参数。将人工测量方法与本文方法进行线性回归分析,得出菌丝体半径、周长、面积和覆盖度的决定系数分别为0.979 5、0.991 5、0.975 0和0.975 0,均方根误差分别为2.20 mm、4.73 mm、176.74 mm^(2)和3.16%。经测试,本文方法能准确地完成食用菌菌丝体表型参数自动测量任务,为食用菌表型分析研究提供理论基础。

基于注意力机制的轻量化VGG玉米籽粒图像识别模型

玉米是重要的生产资料,为实现对玉米种子的识别与保护,实验采集了5个玉米品种,经处理后共获得1778张玉米籽粒图像,建立胚面与胚乳面混合的数据集。按7∶2∶1的比例划分训练集、验证集和测试集。首先基于迁移学习选取DenseNet121、MobileNetV2、VGG16和GoogLeNet对玉米籽粒图像进行识别,在测试集上的准确率分别是94.32%、93.18%、95.45%和92.61%,由于在VGG16上的准确率最高,所以选择对VGG16进行改进,在对模型进行轻量化处理的同时引入通道注意力SE模块,构建一个新的网络模型L-SE-VGG,并与未预训练的VGG16、迁移学习的VGG16和不加SE模块的L-VGG进行对比,最终在L-SE-VGG上的识别准确率高达98.86%。研究为深度学习技术在玉米籽粒品种识别中的应用提供了新的有效策略和实验方法,为玉米籽粒品种的识别和检测提供了参考。

基于VGG-Net的X射线全脊柱冠状面图像分割方法

在计算机辅助脊柱图像分析和疾病诊断应用中,从X射线脊柱图像中自动分割脊柱和椎骨是一个关键且具有挑战性的问题。为进一步提升脊柱图像分割精度,提出一种基于VGG-Net改进的模型。首先,将VGG16网络去掉了后面的全连接层,用作U-Net的特征提取网络;其次,为了增强图像的细节信息,在特征提取网络引入小波分解模块;最后,在上采样网络中设计了一种逐像素相减的自空间注意力模块(SUB-SSAM)机制,进一步提高网络模型识别关键特征的能力。实验结果表明,改进后的模型相较于原VGG-Net模型在平均交并比(mIoU)上提高了2.39%、召回率(recall)提高了0.96%、准确率(accuracy)提高了1.31%,训练的该网络模型可以定位到每一块椎骨,准确分割椎体区域。

基于改进VGG13的冲压件表面缺陷识别方法研究

针对现有冲压件制品缺陷检测方法准确率低的问题,分析深度学习的原理与方法,以VGG13网络为基准模型,通过在特征提取层之后增加CBAM模块进行改进,提出5种基于VGG13与CBAM注意力机制模块相结合的网络模型(VGG13-CBAM),将改进后的新模型与改进前原VGG13模型分别在武汉某制造车间采集的冲压件缺陷数据集上进行实验研究。将数据集以6∶2∶2划分为训练集、验证集、测试集,并使用数据增强进一步扩充训练集,增加模型泛化性能,对比数据增强前后效果的提升。实验结果表明:在改进后的VGG13-CBAM03网络与VGG13-CBAM04网络上效果明显提升,测试集正确率由79.65%分别提高到了81.55%和81.40%,在使用数据增强对训练集进行扩充后,测试集正确率分别达到84.25%和84.15%,有效提升了冲压件缺陷检测准确率。

基于SVDD与VGG的纽扣表面缺陷检测

为解决纽扣表面缺陷检测中人工效率低下,且无需对纽扣表面瑕疵进行分类的问题,提出一种基于DEEP SVDD与改进VGG16的纽扣表面缺陷检测模型。在VGG16中增加BN层加快网络收敛;为提升网络特征提取能力引入SE注意力模块;使用全局平局池化替代全连接层,减少模型参数量,使模型更加健壮。实验结果表明,改进后的模型在DEEP SVDD中的两种方法软边界及一类方法的AUC值分别提升7.7%、5.9%,均高于96%,单张检测时间仅4.5 ms,模型性能满足实际要求。

基于VGG⁃19和MMD卷积神经网络模型的国画风格迁移

卷积神经网络因效果强大而被广泛应用于图像识别,在提取图像特征方面有极大的进步。由于风格迁移技术主要是针对西方油画,而国画是一种传统的中国艺术风格,其在风格迁移方向上缺乏广泛的应用。设计以国画代替西方油画作为风格图像,以自然景观照片作为内容图像,探究传统国画经过卷积神经网络后的提取效果。实验依据VGG算法模型并结合TensorFlow 2框架,对采集的数据集进行预处理,采集像素制成数据矩阵,输入VGG⁃19浅层模型进行训练,通过MMD最小化分布特征图差异,增强卷积层的目标效果。该方法取得比较满意的结果,可为风格迁移转换的研究提供更多参考。

基于复剪切波变换与VGG19模型的医学图像融合方法

针对传统医学图像融合方法存在的细节信息不够清晰、边缘信息易丢失和图像失真等缺点,以及深度学习网络缺乏足够的训练数据集等问题,提出了一种基于复剪切波变换和预训练网络模型VGG19的多模态医学图像融合方法。首先,利用复剪切波变换提取医学图像边缘和纹理信息,并得到多尺度、多方向的子带系数。然后,使用加权局部能量和修正的拉普拉斯算子对低频子带系数进行融合;引入预训练的VGG19提取多层特征图,结合加权评估规则来获取高频子带的融合结果。最后,对融合的高频和低频子带,施加复剪切波逆变换重构融合图像。实验表明,该方法得到的融合图像,不仅可以清晰地显示图像的细节信息和边缘信息,而且能够有效抑制伪影和失真现象的产生,在主观视觉比较和6种客观评价指标下能够达到更佳融合效果。

Automatic Finding of Brain-Tumour Group Using CNN Segmentation and Moth-Flame-Algorithm,Selected Deep and Handcrafted Features

Augmentation of abnormal cells in the brain causes brain tumor(BT),and early screening and treatmentwill reduce its harshness in patients.BT’s clinical level screening is usually performed with Magnetic Resonance Imaging(MRI)due to its multi-modality nature.The overall aims of the study is to introduce,test and verify an advanced image processing technique with algorithms to automatically extract tumour sections from brain MRI scans,facilitating improved accuracy.The research intends to devise a reliable framework for detecting the BT region in the twodimensional(2D)MRI slice,and identifying its class with improved accuracy.The methodology for the devised framework comprises the phases of:(i)Collection and resizing of images,(ii)Implementation and Segmentation of Convolutional Neural Network(CNN),(iii)Deep feature extraction,(iv)Handcrafted feature extraction,(v)Moth-Flame-Algorithm(MFA)supported feature reduction,and(vi)Performance evaluation.This study utilized clinical-grade brain MRI of BRATS and TCIA datasets for the investigation.This framework segments detected the glioma(low/high grade)and glioblastoma class BT.This work helped to get a segmentation accuracy of over 98%with VGG-UNet and a classification accuracy of over 98%with the VGG16 scheme.This study has confirmed that the implemented framework is very efficient in detecting the BT in MRI slices with/without the skull section.

A Large-Scale Group Decision Making Model Based on Trust Relationship and Social Network Updating

With the development of big data and social computing,large-scale group decisionmaking(LGDM)is nowmerging with social networks.Using social network analysis(SNA),this study proposes an LGDM consensus model that considers the trust relationship among decisionmakers(DMs).In the process of consensusmeasurement:the social network is constructed according to the social relationship among DMs,and the Louvain method is introduced to classify social networks to form subgroups.In this study,the weights of each decision maker and each subgroup are computed by comprehensive network weights and trust weights.In the process of consensus improvement:A feedback mechanism with four identification and two direction rules is designed to guide the consensus of the improvement process.Based on the trust relationship among DMs,the preferences are modified,and the corresponding social network is updated to accelerate the consensus.Compared with the previous research,the proposedmodel not only allows the subgroups to be reconstructed and updated during the adjustment process,but also improves the accuracy of the adjustment by the feedbackmechanism.Finally,an example analysis is conducted to verify the effectiveness and flexibility of the proposed method.Moreover,compared with previous studies,the superiority of the proposed method in solving the LGDM problem is highlighted.

Introducing hydroxyl groups to tailor the d-band center of Ir atom through side anchoring for boosted ORR and HER

Tuning the coordination atoms of central metal is an effective means to improve the electrocatalytic activity of atomic catalysts.Herein,iridium(Ir) is proposed to be asymmetrically anchored by sp-N and pyridinic N of hydrogen-substituted graphdiyne(HsGDY),and coordinated with OH as an Ir atomic catalyst(Ir_(1)-N-HsGDY).The electron structures,especially the d-band center of Ir atom,are optimized by these specific coordination atoms.Thus,the as-synthesized Ir_(1)-N-HsGDY exhibits excellent electrocatalytic performances for oxygen reduction and hydrogen evolution reactions in both acidic and alkaline media.Benefiting from the unique structure of HsGDY,IrN_(2)(OH)_(3) has been developed and demonstrated to act as the active site in these electrochemical reactions.All those indicate the fresh role of the sp-N in graphdiyne in producing a new anchor way and contributing to promote the electrocatalytic activity,showing a new strategy to design novel electrochemical catalysts.

End-group modulation of phenazine based non-fullerene acceptors for efficient organic solar cells with high open-circuit voltage

Phenazine-based non-fullerene acceptors(NFAs)have demonstrated great potential in improving the power conversion efficiency(PCE)of organic solar cells(OSCs).Halogenation is known to be an effective strategy for increasing optical absorption,refining energy levels,and improving molecular packing in organic semiconductors.Herein,a series of NFAs(Pz IC-4H,Pz IC-4F,Pz IC-4Cl,Pz IC-2Br)with phenazine as the central core and with/without halogen-substituted(dicyanomethylidene)-indan-1-one(IC)as the electron-accepting end group were synthesized,and the effect of end group matched phenazine central unit on the photovoltaic performance was systematically studied.Synergetic photophysical and morphological analyses revealed that the PM6:Pz IC-4F blend involves efficient exciton dissociation,higher charge collection and transfer rates,better crystallinity,and optimal phase separation.Therefore,OSCs based on PM6:Pz IC-4F as the active layer exhibited a PCE of 16.48%with an open circuit voltage(Voc)and energy loss of 0.880 V and 0.53 e V,respectively.Accordingly,this work demonstrated a promising approach by designing phenazine-based NFAs for achieving high-performance OSCs.

A Novel On-Site-Real-Time Method for Identifying Characteristic Parameters Using Ultrasonic Echo Groups and Neural Network

On-site and real-time non-destructive measurement of elastic constants for materials of a component in a in-service structure is a challenge due to structural complexities,such as ambiguous boundary,variable thickness,nonuniform material properties.This work develops for the first time a method that uses ultrasound echo groups and artificial neural network(ANN)for reliable on-site real-time identification of material parameters.The use of echo groups allows the use of lower frequencies,and hence more accommodative to structural complexity.To train the ANNs,a numerical model is established that is capable of computing the waveform of ultrasonic echo groups for any given set of material properties of a given structure.The waveform of an ultrasonic echo groups at an interest location on the surface the structure with material parameters varying in a predefined range are then computed using the numerical model.This results in a set of dataset for training the ANN model.Once the ANN is trained,the material parameters can be identified simultaneously using the actual measured echo waveform as input to the ANN.Intensive tests have been conducted both numerically and experimentally to evaluate the effectiveness and accuracy of the currently proposed method.The results show that the maximum identification error of numerical example is less than 2%,and the maximum identification error of experimental test is less than 7%.Compared with currently prevailing methods and equipment,the proposefy the density and thickness,in addition to the elastic constants.Moreover,the reliability and accuracy of inverse prediction is significantly improved.Thus,it has broad applications and enables real-time field measurements,which has not been fulfilled by any other available methods or equipment.

Channel Correlation Based User Grouping Algorithm for Nonlinear Precoding Satellite Communication System

Low Earth Orbit(LEO)multibeam satellites will be widely used in the next generation of satellite communication systems,whose inter-beam interference will inevitably limit the performance of the whole system.Nonlinear precoding such as Tomlinson-Harashima precoding(THP)algorithm has been proved to be a promising technology to solve this problem,which has smaller noise amplification effect compared with linear precoding.However,the similarity of different user channels(defined as channel correlation)will degrade the performance of THP algorithm.In this paper,we qualitatively analyze the inter-beam interference in the whole process of LEO satellite over a specific coverage area,and the impact of channel correlation on Signal-to-Noise Ratio(SNR)of receivers when THP is applied.One user grouping algorithm is proposed based on the analysis of channel correlation,which could decrease the number of users with high channel correlation in each precoding group,thus improve the performance of THP.Furthermore,our algorithm is designed under the premise of co-frequency deployment and orthogonal frequency division multiplexing(OFDM),which leads to more users under severe inter-beam interference compared to the existing research on geostationary orbit satellites broadcasting systems.Simulation results show that the proposed user grouping algorithm possesses higher channel capacity and better bit error rate(BER)performance in high SNR conditions relative to existing works.

Molecular packing tuning via chlorinated end group enables efficient binary organic solar cells over 18.5%

Designing novel nonfullerene acceptors(NFAs)is of vital importance for the development of organic solar cells(OSC).Modification on the side chain and end group are two powerful tools to construct efficient NFAs.Here,based on the high-performance L8BO,we selected 3-ethylheptyl to substitute the inner chain of 2-ethylhexyl,obtaining the backbone of BON3.Then we introduced different halogen atoms of fluorine and chlorine on 2-(3-oxo-2,3-dihydro-1Hinden-1-ylidene)malononitrile end group(EG)to construct efficient NFAs named BON3-F and BON3-Cl,respectively.Polymer donor D18 was chosen to combine with two novel NFAs to construct OSC devices.Impressively,D18:BON3-Cl-based device shows a remarkable power conversion efficiency(PCE)of 18.57%,with a high open-circuit voltage(V_(OC))of 0.907 V and an excellent fill factor(FF)of 80.44%,which is one of the highest binary PCE of devices based on D18 as the donor.However,BON3-F-based device shows a relatively lower PCE of 17.79%with a decreased FF of 79.05%.The better photovoltaic performance is mainly attributed to the red-shifted absorption,higher electron and hole mobilities,reduced charge recombination,and enhanced molecular packing in the D18:BON3-Cl films.Also,we performed stability tests on two binary systems;the D18:BON3-Cl and D18:BON3-F devices maintain 88.1%and 85.5%of their initial efficiencies after 169 h of storage at 85°C in an N2-filled glove box,respectively.Our work demonstrates the importance of selecting halogen atoms on EG and provides an efficient binary system of D18:BON3-Cl for further improvement of PCE.

E2E-MFERC:AMulti-Face Expression Recognition Model for Group Emotion Assessment

In smart classrooms, conducting multi-face expression recognition based on existing hardware devices to assessstudents’ group emotions can provide educators with more comprehensive and intuitive classroom effect analysis,thereby continuouslypromotingthe improvementof teaching quality.However,most existingmulti-face expressionrecognition methods adopt a multi-stage approach, with an overall complex process, poor real-time performance,and insufficient generalization ability. In addition, the existing facial expression datasets are mostly single faceimages, which are of low quality and lack specificity, also restricting the development of this research. This paperaims to propose an end-to-end high-performance multi-face expression recognition algorithm model suitable forsmart classrooms, construct a high-quality multi-face expression dataset to support algorithm research, and applythe model to group emotion assessment to expand its application value. To this end, we propose an end-to-endmulti-face expression recognition algorithm model for smart classrooms (E2E-MFERC). In order to provide highqualityand highly targeted data support for model research, we constructed a multi-face expression dataset inreal classrooms (MFED), containing 2,385 images and a total of 18,712 expression labels, collected from smartclassrooms. In constructing E2E-MFERC, by introducing Re-parameterization visual geometry group (RepVGG)block and symmetric positive definite convolution (SPD-Conv) modules to enhance representational capability;combined with the cross stage partial network fusion module optimized by attention mechanism (C2f_Attention),it strengthens the ability to extract key information;adopts asymptotic feature pyramid network (AFPN) featurefusion tailored to classroomscenes and optimizes the head prediction output size;achieves high-performance endto-end multi-face expression detection. Finally, we apply the model to smart classroom group emotion assessmentand provide design references for classroom effect analysis evaluation metrics. Experiments based on MFED showthat the mAP and F1-score of E2E-MFERC on classroom evaluation data reach 83.6% and 0.77, respectively,improving the mAP of same-scale You Only Look Once version 5 (YOLOv5) and You Only Look Once version8 (YOLOv8) by 6.8% and 2.5%, respectively, and the F1-score by 0.06 and 0.04, respectively. E2E-MFERC modelhas obvious advantages in both detection speed and accuracy, which can meet the practical needs of real-timemulti-face expression analysis in classrooms, and serve the application of teaching effect assessment very well.

How Group Leaders Build Stable Community Buying Groups:A Perspective Based on the Differential Mode of Association

Since 2016,community group buying has grown significantly in China,largely driven by its efficient logistics,supply chains,low prices,and convenience.This model has been further popularized during the COVID-19 pandemic due to its effectiveness in meeting daily needs while minimizing human-to-human contact.A key component of this business model is the“group leaders”-influential individuals within a community responsible for managing group buying activities,which include order collection,supplier liaison,and goods distribution.Their primary task is to form and sustain a reliable community group buying consortium,a task that demands excellent organizational and interpersonal skills.This paper examines this phenomenon using the lens of the differential mode of association,a theoretical model explaining interpersonal relationships in traditional Chinese society.The research indicates that group leaders,through regular interaction with consumers,are able to alter their social network position,increase their influence,understand consumer needs,provide satisfying services,and enhance trust,thereby transforming consumers into loyal group buying participants.This transformation not only brings stability to group buying activities but also reinforces the community influence of group leaders,thus fostering the growth of community group buying.

Spherical Functions on Fuzzy Lie Group

Let G be a locally compact Lie group and its Lie algebra. We consider a fuzzy analogue of G, denoted by called a fuzzy Lie group. Spherical functions on are constructed and a version of the existence result of the Helgason-spherical function on G is then established on .

Combined Promoting Effects of Specific Organic Functional Groups and Alumina Surface Characteristics for the Design of a Highly Efficient NiMo/Al_(2)O_(3) Hydrodesulfurization Catalyst

To prepare a highly efficient NiMo/Al_(2)O_(3) hydrodesulfurization catalyst,the combined effects of specific organic functional groups and alumina surface characteristics were investigated.First,the correlation between the surface characteristics of four different alumina and the existing Mo species states was established.It was found that the Mo equilibrium adsorption capacity can be used as a specific descriptor to quantitatively evaluate the changes in surface characteristics of different alumina.A lower Mo equilibrium adsorption capacity for alumina means weaker metal-support interaction and the loaded Mo species are easier to transform into MoS2.However,the Mo-O-Al bonds still exist at the metal-support interface.The introduction of cationic surfactant hecadecyl trimethyl ammonium bromide(CTAB)can further improve Mo species dispersion through electrostatic attraction with Mo anions and interaction of its alkyl chain with the alumina surface;meanwhile,the introduction of ethylenediamine tetraacetic acid(EDTA)can complex with Ni ions to enhance the Ni-promoting effect on Mo.Therefore,the NiMo catalyst designed using alumina with lower Mo equilibrium adsorption capacity and the simultaneous addition of EDTA and CTAB exhibits the highest hydrodesulfurization activity for 4,6-dimethyl dibenzothiophene because of its proper metal-support interaction and more well-dispersed Ni-Mo-S active phases.