融合内外部特征水印的模型保护方案

针对经典模型水印技术在保护模型所有权过程中存在鲁棒性差、提取率低等问题,融合白、黑盒水印优势,提出了一种特征嵌入的模型保护方案。按照香农熵大小进行数据集样本划分的策略,将数据集样本划分为良性样本、风格迁移样本、关键密钥样本;利用风格迁移样本集对模型嵌入外部特征,将关键密钥样本标签嵌入模型内部特征;通过训练二元分类器并利用掩码梯度下降方法修改极少量参数让模型产生特定输出来综合判断模型是否被窃取。实验结果表明,所提方案用较小开销保证了水印的高保真度,在标签查询、知识蒸馏等攻击下仍具有较高稳定性,且能规避恶意检测风险。

Comparing the efficacy and safety of low,medium,and high dosages of selexipag for treating pulmonary hypertension:A systematic review and meta-analysis

Background:The maintenance dosage of selexipag is categorized as low,medium or high.In order to assess the efficacy and safety of different dosages of selexipag for the risk stratification of pulmonary arterial hypertension(PAH),we performed a sys-tematic review and meta-analysis.Methods:Studies assessing PAH risk stratification indices,such as the World Health Organization functional class(WHO-FC),six-minute walk distance(6MWD),N-terminal pro-B-type natriuretic peptide(NT-proBNP)level,right atrial pressure(RAP),cardiac index(CI)and mixed venous oxygen saturation(SvO2),were included.Results:Thirteen studies were included.Selexipag led to improvements in the 6MWD(MD:24.20 m,95%CI:10.74-37.67),NT-proBNP(SMD:-0.41,95%CI:-0.79-0.04),CI(MD:0.47 L/min/m^(2),95%CI:0.17-0.77)and WHO-FC(OR:0.564,95%CI:0.457-0.697).Subgroup analysis demonstrated that all three dosages improved the 6MWD.A moderate dosage led to improvements in the CI(MD:0.30 L/min/m^(2),95%CI:0.15-0.46)and WHO-FC(OR:0.589,95%CI:0.376-0.922).Within 6 months of treatment,only the WHO-FC and CI were significantly improved(OR:0.614,95%CI:0.380-0.993;MD:0.30 L/min/m^(2),95%CI:0.16-0.45,respectively).More than 6 months of treatment significantly improved the 6MWD,WHO-FC and NT-proBNP(MD:40.87 m,95%CI:10.97-70.77;OR:0.557,95%CI:0.440-0.705;SMD:-0.61,95%CI:-1.17-0.05,respectively).Conclusions:Low,medium,and high dosages of selexipag all exhibited good effects.When treatment lasted for more than 6 months,selexipag exerted obvious effects,even in the low-dosage group.This finding is important for guiding individualized treatments.

CircPMS1 promotes proliferation of pulmonary artery smooth muscle cells,pulmonary microvascular endothelial cells,and pericytes under hypoxia

Background:Circular RNAs(circRNAs)have been recognized as significant regulators of pulmonary hypertension(PH);however,the differential expression and function of circRNAs in different vascular cells under hypoxia remain unknown.Here,we identified co-differentially expressed circRNAs and determined their putative roles in the proliferation of pulmonary artery smooth muscle cells(PASMCs),pulmonary microvascular endothelial cells(PMECs),and pericytes(PCs)under hypoxia.Methods:Whole transcriptome sequencing was performed to analyze the differential expression of circRNAs in three different vascular cell types.Bioinformatic analysis was used to predict their putative biological function.Quantitative real-time polymerase chain reaction,Cell Counting Kit-8,and EdU Cell Proliferation assays were carried out to determine the role of circular postmeiotic segregation 1(circPMS1)as well as its potential sponge mechanism in PASMCs,PMECs,and PCs.Results:PASMCs,PMECs,and PCs exhibited 16,99,and 31 differentially expressed circRNAs under hypoxia,respectively.CircPMS1 was upregulated in PASMCs,PMECs,and PCs under hypoxia and enhanced the proliferation of vascular cells.CircPMS1may upregulate DEP domain containing 1(DEPDC1)and RNA polymerase II subunit D expression by targeting microRNA-432-5p(miR-432-5p)in PASMCs,upregulate MAX interactor 1(MXI1)expression by targeting miR-433-3p in PMECs,and upregulate zinc finger AN1-type containing 5(ZFAND5)expression by targeting miR-3613-5p in PCs.Conclusions:Our results suggest that circPMS1 promotes cell proliferation through the miR-432-5p/DEPDC1 or miR-432-5p/POL2D axis in PASMCs,through the miR-433-3p/MXI1 axis in PMECs,and through the miR-3613-5p/ZFAND5 axis in PCs,which provides putative targets for the early diagnosis and treatment of PH.

Neural stem cell-derived exosomes regulate cell proliferation,migration,and cell death of brain microvascular endothelial cells via the miR-9/Hes1 axis under hypoxia

Background:Our previous study found that mouse embryonic neural stem cell(NSC)-derived exosomes(EXOs)regulated NSC differentiation via the miR-9/Hes1 axis.However,the effects of EXOs on brain microvascular endothelial cell(BMEC)dysfunction via the miR-9/Hes1 axis remain unknown.Therefore,the current study aimed to determine the effects of EXOs on BMEC proliferation,migration,and death via the miR-9/Hes1 axis.Methods:Immunofluorescence,quantitative real-time polymerase chain reaction,cell counting kit-8 assay,wound healing assay,calcein-acetoxymethyl/propidium iodide staining,and hematoxylin and eosin staining were used to determine the role and mechanism of EXOs on BMECs.Results:EXOs promoted BMEC proliferation and migration and reduced cell death under hypoxic conditions.The overexpression of miR-9 promoted BMEC prolifera-tion and migration and reduced cell death under hypoxic conditions.Moreover,miR-9 downregulation inhibited BMEC proliferation and migration and also promoted cell death.Hes1 silencing ameliorated the effect of amtagomiR-9 on BMEC proliferation and migration and cell death.Hyperemic structures were observed in the regions of the hippocampus and cortex in hypoxia-induced mice.Meanwhile,EXO treatment improved cerebrovascular alterations.Conclusion:NSC-derived EXOs can promote BMEC proliferation and migra-tion and reduce cell death via the miR-9/Hes1 axis under hypoxic conditions.Therefore,EXO therapeutic strategies could be considered for hypoxia-induced vascular injury.

溶剂分离法提取生物柴油中的不饱和脂肪酸甲酯

为了降低生物柴油的冷凝点,采用溶剂分离法提取生物柴油中的不饱和脂肪酸甲酯,重点考察了醇的种类、溶剂的用量、冷冻温度和过滤方式对分离效果的影响。结果表明:乙醇的用量为原料质量的2~3倍,冷冻温度-15℃,压滤法过滤,最高可得到96.9575%的不饱和脂肪酸甲酯含量和86.92%的得率。

Study on the mechanism of SND in the treatment of anxiety disorders based on network pharmacology, molecular docking and experiment validation

Background:Sini decoction(SND)is a classic traditional Chinese medicine(TCM)formulation that can be used to treat anxiety-related disorders,but the active substance and underlying molecular mechanism of its anxiolytic effects are unknown.In this study,network pharmacology,molecular docking research and experimental verification methods were used to preliminarily explore the bioactive compounds and potential target mechanisms of SND anxiolytic.Methods:The active components and corresponding targets of SND were collected by TCMSP.GeneCards,OMIM,PharmGkb,TTD and Drugbank were used to search for the targets of anxiety disorders.The core target of SND in the treatment of anxiety was screened by PPI.R language was used to analyze the intersection targets of SND in the treatment of anxiety disorders by GO and KEGG enrichment analysis.AutoDock Vina was used for molecular docking,and Discovery Studio was used for visual conformation analysis after docking.The anti-anxiety effect and molecular mechanism of SND were studied by in vivo experiment.Results:Based on network pharmacological analysis,we obtained 112 active ingredients and 350 effective targets related to anxiety from SND.In PPI analysis,26 targets such as STAT3,MAPK3,MAPK1,MAPK14,SRC,HSP90AA1,TP53 and PIK3CA were identified as core targets.GO and KEGG analysis showed that the anxiolytic mechanism of SND may be related to the neuroactive ligand-receptor interaction pathway and inflammatory pathway.Molecular docking showed that quercetin,naringenin,licochalcone A had high affinity with JAK2,MAPK14 and MAPK3.Animal experiments have shown that SND reverses the upregulation of GluN2B(NMDAR)and GluA1(AMPAR)proteins,and SND improves anxiety disorders by regulating glutamate transmitter levels,which may be related to neuroactive ligand-receptor interaction pathways,particularly glutamate receptors.Conclusion:This study shows that SND can improve FS-induced behavioral changes in mice and can modulate hippocampal synapse-associated protein defects,partially reversing glutamate receptor expression through the neuroactive ligand-receptor interaction pathway,and further improved anxiety disorders.At the same time,combined with network pharmacology and molecular docking,the key components,core targets and related pathways of SND are discussed,which shows that the active components of SND play an effective role in anxiety through multi-targets and multi-pathways,which provides a reference for the material basis and mechanism of SND.

基于最小数据集的宁都县不同整地方式下果园土壤质量评价

为探究不同整地措施下梯田果园土壤质量的变化,以赣南宁都县小洋小流域脐橙果园开发示范区内的3种典型坡面(优化整地T y、传统整地T c、撂荒地CK)为研究对象,采用主成分分析法(principal component analysis,PCA)筛选最小数据集指标,对该地区土壤质量进行评价。结果表明:(1)由全量数据集(total data set,TDS)和最小数据数据集(minimum data set,MDS)计算的土壤质量指数存在显著正相关关系,R^(2)=0.604,表明MDS可以代替TDS对宁都县脐橙果园开发示范区进行土壤质量评价;(2)3种典型坡面的土壤质量指数(soil quality index,SQI)表现为优化整地(SQI=0.477)>撂荒地(SQI=0.436)>传统整地(SQI=0.220),其中T y与CK之间无显著差异,T y的SQI相较于T c提升1.16倍,表明优化整地措施下土壤质量有显著提升;(3)果园不同土层总体表现为表层(0—10 cm)土壤质量指数大于深层(10—20 cm)土壤,通过土壤质量指数分级发现,优化整地梯田和撂荒地土壤质量优于传统整地。研究结果可以为我国红壤丘陵区的坡地资源利用管理以及梯田修建后水土保持措施布设提供科学依据。

RFID安全认证协议综述

在物联网发展中,RFID技术以其轻量化的优势在物联网体系中占据重要地位。同时,RFID安全认证协议也因物理条件限制受到安全威胁。首先,通过对现行主流RFID安全认证协议进行梳理,按加密算法的量级将其划分为超轻量级、轻量级、中量级和重量级安全认证协议;然后,对其中典型的安全认证协议存在的安全问题进行分析,对近年来提出的改进协议安全性能及性能指标按量级进行讨论比较;最后,探讨了RFID安全认证协议可能的发展方向。

丹江口库区典型小流域生物结皮驱动土壤性质变化及分离响应

为探究生物结皮发育驱动下土壤水蚀特征及机理,选取苔藓发育为优势种的退耕样地,设计5个盖度等级(1%~20%,20%~40%,40%~60%,60%~80%,80%~100%),以裸地为对照,在不同水流剪切力(6.12~19.64 Pa)下进行冲刷试验。结果表明:结皮发育显著影响土壤养分含量,且高盖度结皮覆盖下养分表聚效应更加明显。生物结皮发育显著影响土壤分离能力,裸地土壤分离能力[1.4656 kg/(m^(2)·s)]为生物结皮[0.0042~0.4682 kg/(m^(2)·s)]的3.58~348.95倍,裸地细沟可蚀性(0.2935 s/m)为生物结皮(0.0010~0.0822 s/m)的3.57~293.50倍。相对土壤分离能力及细沟可蚀性均随结皮盖度增加呈指数函数衰减,结皮冲破时间随结皮盖度增加呈指数函数增长;非线性回归分析表明,土壤分离能力可用水流剪切力、生物结皮盖度和土壤黏结力进行模拟(NSE=0.749)。综上,生物结皮发育显著提升土壤抗侵蚀性能,研究结果可为半湿润区水土流失防治及生物结皮管理利用提供理论参考。

红壤丘陵区不同盖度生物结皮对水分入渗的影响

为探究不同盖度生物结皮对入渗特征的影响,在红壤丘陵区咸宁选取第四纪黏土发育土壤(S型)和泥质页岩发育土壤(N型)上的生物结皮,设计6个盖度水平(裸地,1%~20%,20%~40%,40%~60%,60%~80%,80%~100%),使用微型圆盘入渗仪(mini disk)测定土壤入渗过程,探究入渗过程的影响因素,同时采用3种常见的入渗模型对入渗过程进行拟合,并评价模型适用性。结果表明:(1)相对于裸地,生物结皮发育提高土壤有机质和黏粒含量,增加结皮厚度和生物量,降低砂粒含量。(2)第四纪黏土发育土壤水分入渗特征随生物结皮盖度的增加而逐渐降低,初始入渗速率、稳定入渗速率、平均入渗速率、非饱和导水率的变化范围分别为0.25~1.55,0.13~0.91,0.17~1.11,2.04~8.48 mm/min。泥质页岩发育土壤除40%~60%盖度砾石含量较高,土壤水分入渗也随盖度的增加而逐渐减小,80%~100%盖度的初始、平均和稳定入渗率较裸地分别降低91.14%,87.64%和91.30%,生物结皮的存在阻碍土壤水分入渗。冗余分析表明,对于第四纪黏土发育土壤,生物量(54.30%)对其入渗特征的解释度最高,泥质页岩发育土壤则为盖度(39.30%)和砾石含量(34.00%)。(3)Horton模型、Kostiakov模型、Philip模型3种主流入渗模型中Horton模型拟合效果最优。研究结果可为揭示湿润区生物结皮水分运移规律提供科学参考。

基于土壤侵蚀时空演变的县域生态承载力评价

[目的]探究土地类型变化背景下土壤侵蚀和生态承载力的时空演变及相关关系,对提高生态承载力水平、优化土地利用结构和水土保持措施配置具有重要意义。[方法]以江西省宁都县为研究区域,基于自然资源和社会经济数据,运用InVEST模型与空间主成分方法,分析了土壤侵蚀变化对生态承载力的影响。[结果](1)宁都县主要土地利用类型为林地、耕地和草地,9年来土壤侵蚀程度2010年>2015年>2018年,分别为597.42,591.29,583.51 t/(km^(2)·a),侵蚀类型以微度为主,在中部、西北部和东南部等地区,建设用地和耕地土壤侵蚀程度严重。(2)宁都县生态承载力西南低、东北高,9年来生态承载力整体上略有好转;三期土壤侵蚀程度严重的区域,生态承载力水平低。(3)2010—2018年,宁都县开展规模化整地,使林地、荒地(未利用地)转为园地、耕地,增加了土地覆被度,加之降雨量减小,径流冲刷减弱,侵蚀模数大幅降低,分别减少了178.19,876.32,2205.07 t/(km^(2)·a),其他土地利用类型侵蚀模数变化较小,总体来看土壤侵蚀模数变化的主要原因是耕地、园地和未利用地的转化。[结论]宁都县不同土地利用方式下土壤侵蚀强度不同,且对区域承载力有较大影响,未来应加强人类活动对侵蚀及生态承载力的影响研究。

Defect Engineering of Carbons for Energy Conversion and Storage Applications

Sustainable energy conversion and storage technologies are a vital prerequisite for neutral future carbon.To this end,carbon materials with attractive features,such as tunable pore architecture,good electrical conductivity,outstanding physicochemical stability,abundant resource,and low cost,have used as promising electrode materials for energy conversion and storage.Defect engineering could modulate the structures of carbon materials,thereby affecting their electronic properties.The presence of defects on carbons may lead to asymmetric charge distribution,change in geometrical configuration,and distortion of the electronic structure that may result in unexpected electrochemical performances.In this review,recent advances in defects of carbons used for energy conversion and storage were examined in terms of types,regulation strategies,and fine characterization means of defects.The applications of such carbons in supercapacitors,rechargeable batteries,and electrocatalysis were also discussed.The perspectives toward the development of defect engineering carbons were proposed.In all,novel insights related to improvement in high-performance carbon materials for future energy conversion and storage applications were provided.

基于R-FCN区域全卷积网络的绝缘子红外图像识别研究

红外热成像因其具有非接触性、灵敏性等优点,已被广泛应用于电力设备的带电检测及其诊断中。其中,对设备快速精确地识别定位是电力设备智能诊断的关键。然而利用传统机器算法对电力设备图像进行识别定位,存在泛化能力不强、鲁棒性较差等不足。针对此问题,开展基于R-FCN区域全卷积网络的绝缘子红外图像识别研究。在TensorFlow框架下搭建R-FCN检测模型,并利用迁移学习方法初始化模型权重,以提高训练效果。最后,将所研究算法与Faster-RCNN和SSD模型进行对比。实验表明,R-FCN模型的检测精度为89.2%,检测速度为23 fps,具有较高的精度和速度。该算法为绝缘子的智能诊断奠定坚实基础。

Biology and function of pericytes in the vascular microcirculation

Pericytes are the main cellular components of tiny arteries and capillaries.Studies have found that pericytes can undergo morphological contraction or relaxation under stimulation by cytokines,thus affecting the contraction and relaxation of microvessels and playing an essential role in regulating vascular microcirculation.Moreover,due to the characteristics of stem cells,pericytes can differentiate into a variety of inflammatory cell phenotypes,which then affect the immune function.Additionally,pericytes can also participate in angiogenesis and wound healing by interacting with endothelial cells in vascular microcirculation disorders.Here we review the origin,biological phenotype and function of pericytes,and discuss the potential mechanisms of pericytes in vascular microcirculation disorders,especially in pulmonary hypertension,so as to provide a sound basis and direction for the prevention and treatment of vascular microcirculation diseases.

基于虚拟自博弈多智能体近端优化策略的无人机对抗决策

研究了基于虚拟自博弈多智能体近端策略优化的无人机对抗决策问题。无人机对抗依赖自主决策,使无人机能够根据环境信息生成行动指令。提出了一种基于红蓝空战任务的无人机对抗自主决策方法。首先,采用导弹攻击区域与无人机之间的相对角度来评估当前情况。然后,以场景评估为指导,进行状态空间、动作空间和实时奖励反馈的设计,简化训练过程。在此基础上,提出了一种利用虚拟自博弈多智能体近端策略的方法,旨在从训练数据的经验缓冲区中推导出优势函数和平均策略。最后,通过对无人机执行红蓝对抗任务的仿真,验证了该方法的有效性和优势所在。

南盘江盆地板街古油藏流体包裹体特征及其成藏成矿关系研究

在南盘江盆地发现了过演化程度(高成熟—过成熟)的烃类包裹体,其主要赋存于与板街古油藏储层沥青共生的方解石脉中。通过镜下鉴定、激光拉曼测试、显微测温等手段,对烃类包裹体和与其同期的盐水包裹体进行岩相学和地球化学研究。结果表明,储层中包裹体分为盐水包裹体和有机包裹体两大类,有机包裹体又进一步划分为气液两相包裹体、气相包裹体和固相沥青包裹体,气相组分主要为甲烷和乙烷,固相组分主要为碳质沥青。盐水包裹体的均一温度为119~150℃,代表了包裹体的捕获温度;而有机包裹体的均一温度有127~161℃和165~184.3℃两个区间,则代表了原先捕获的烃类包裹体在未被破坏的情况下经历热演化的温度区间。该古油藏中包裹体为泥盆系烃源岩在热裂解生湿气和生干气时期原油裂解生成的气烃进入二叠系储层,被早期生成的方解石捕获。保存在包裹体内的烃类经历了热演化,最终形成干气和沥青包裹体,这些包裹体记录了板街古油藏自油气藏演化至干气藏的油气活动事件。下一步应在储层或矿床不同层位进行采样,并进行地球化学分析,对成藏成矿关系进一步研究。

通信网络运维多维评估模型设计及应用

通信网络运行维护工作是电信运营商的一项基础性重要工作。当前,世界各运营商网络运维工作公开披露的信息相对较少,缺乏评估各运营商电信网络运维水平的有效模型。以平衡积分卡方法为基础,结合长期通信网络运维体系研究及工作实践,创新性地提出了通信网络运维多维评估模型,并应用该模型对国际大型电信运营商、创新型电信运营商、高增长型电信运营商3种不同类型的运营商进行对比分析,以期对国内运营商网络运维工作发展提供有效的评估和指引。

伴有SORL1基因突变的阿尔茨海默病1例报道

目的:阿尔茨海默病(AD)是一种神经变性疾病,近年来分拣蛋白相关受体基因(SORL1)作为其风险基因越来越得到重视,现将1例伴有SORL1基因突变的早发型阿尔茨海默病(EOAD)进行报道。方法:收集1例伴有SORL1基因突变的AD患者临床资料,包括病史特点、神经系统查体、神经心理学检查、常规检验、MRI、PET-CT及基因检测,并复习相关文献。结果:该患者49岁起病,以记忆力减退为主,之后计算力等各项认知功能持续性下降,伴肌张力增高、运动迟缓,缺乏联带运动等锥体外系症状,日常生活不能自理;头颅MRI显示脑萎缩,以颞叶、海马萎缩为著;基因检测显示SORL1基因43号外显子c.5744A>G杂合突变(p.Y1915C)。结论:本患者发病年龄,临床表现,影像和脑脊液异常,存在SORL1基因突变,诊断符合SORL1基因突变的EOAD,该病例锥体外系症状有助于丰富SORL1突变相关的EOAD临床特征,对临床诊疗工作具有提示作用。

戏剧影视文学专业研究生培养模式实践探索与研究

更新教学理念,开展多样化教学实践是培养创新型和复合型研究生人才的重要手段,也是人才培养协同提质的途径。以项目带动实践,以专题引领创新,是实现戏剧影视文学专业研究生教育从学术型逐步向专业型转变的有效方法。科学的评价机制是实现人才培养增效保质的重要保障。

中国西北半干旱区净初级生产力驱动因子空间计算分析

净初级生产力(Net Primary Productivity,NPP)驱动因子分析对区域生态系统生产力监控与预测及生态承载力评估有重要意义。NPP驱动力系统研究近年成果已有很多,但前人研究中未能将因子空间属性在原数据和分析中得到体现。实际上,这不利于获得更科学的分析结果,甚至阻碍了数据本身空间信息的表达。在前人研究分析基础上,以新疆伊犁河谷为试验研究区,使用一种全新的数据预处理方法(C.V计算,C.V=SD/Mean),通过该方法使数据固有的空间属性或空间关系信息得以表达。将该数据集与原始数据集统计分析结果对比分析,得出结论:(1)半干旱区环境下NDVI、积温和海拔与NPP相关性相对较强。经自变量因子重要性排序分析,发现以上因子对因变量(NPP)有较高重要性(P<0.01);(2)C.V数据集较原数据集在模型建立中拥有更高的拟合度,r值平均高出0.1左右,结果更优(P<0.01),在环境因子建模分析中具有较好的应用前景;(3)原始数据仅能表达因子在空间中的数值分布,经C.V计算处理后,能直观的表达因子的空间变异性和数组单元的空间紊乱程度。(4)在半干旱区海拔因子对NPP的作用力相对较强,Duncan分析发现,各因子在海拔尺度下存在显著的分异特征(P<0.01),说明了海拔因子对NPP的作用力。综上,研究尝试使用C.V计算数据预处理,令原始数据集附带空间属性,使得研究分析结果不单纯依赖于数值关系,结果表达的关系更加全面。研究结果对生态环境因子分析及NPP驱动力分析研究都有重要的试验价值与科学意义。