基于伪标签的无监督领域自适应行人重识别研究综述

行人重识别是计算机视觉领域的热点研究课题之一。近年来,为了解决行人重识别实际应用中标签数据稀缺的问题,同时也为了有效地利用现有的标签数据,研究者们提出了基于生成对抗网络以及基于伪标签的领域自适应方法,用于进行跨领域的行人重识别研究。基于伪标签的无监督领域自适应行人重识别方法由于效果显著而备受研究者的青睐。文中梳理了近7年来基于伪标签的无监督领域自适应行人重识别的研究成果,将基于伪标签的方法从模型训练角度划分为两个阶段。1)伪标签生成阶段。现有工作的伪标签生成方法大多使用聚类方法,部分工作采用基于图结构学习的图匹配、图卷积网络方法来生成目标域的伪标签。2)伪标签精炼阶段。文中将现有的伪标签精炼方法归纳为基于表征学习的精炼方法以及基于相似度学习的精炼方法,并分别进行模型方法的总结与整理。最后,讨论现阶段基于伪标签的无监督领域自适应行人重识别面临的挑战并对未来可能的发展方向进行展望。

Study of Flow and Heat Transfer in an Ejector-Driven Swirl Anti-Icing Chamber

The formation of ice on the leading edge of aircraft engines is a serious issue,as it can have catastrophic consequences.The Swirl Anti-Icing(SAI)system,driven by ejection,circulates hot fluid within a 360°annular chamber to heat the engine inlet lip surface and prevent icing.This study employs a validated Computational Fluid Dynamics(CFD)approach to study the impact of key geometric parameters of this system on flow and heat transfer characteristics within the anti-icing chamber.Additionally,the entropy generation rate and exergy efficiency are analyzed to assess the energy utilization in the system.The research findings indicate that,within the considered flow range,reducing the nozzle specific areaφfrom 0.03061 to 0.01083 can enhance the ejection coefficient by over 60.7%.This enhancement increases the air circulating rate,thereby intensifying convective heat transfer within the SAI chamber.However,the reduction inφalso leads to a significant increase in the required bleed air pressure and a higher entropy generation rate,indicating lower exergy efficiency.The nozzle angleθnotably affects the distribution of hot and cold spots on the lip surface of the SAI chamber.Increasingθfrom 0°to 20°reduces the maximum temperature difference on the anti-icing chamber surface by 60 K.

Developing and validating a predictive model of delivering large-forgestational-age infants among women with gestational diabetes mellitus

BACKGROUND The birth of large-for-gestational-age(LGA)infants is associated with many shortterm adverse pregnancy outcomes.It has been observed that the proportion of LGA infants born to pregnant women with gestational diabetes mellitus(GDM)is significantly higher than that born to healthy pregnant women.However,traditional methods for the diagnosis of LGA have limitations.Therefore,this study aims to establish a predictive model that can effectively identify women with GDM who are at risk of delivering LGA infants.AIM To develop and validate a nomogram prediction model of delivering LGA infants among pregnant women with GDM,and provide strategies for the effective prevention and timely intervention of LGA.METHODS The multivariable prediction model was developed by carrying out the following steps.First,the variables that were associated with LGA risk in pregnant women with GDM were screened by univariate analyses,for which the P value was<0.10.Subsequently,Least Absolute Shrinkage and Selection Operator regression was fit using ten cross-validations,and the optimal combination factors were se-lected by choosing lambda 1se as the criterion.The final predictors were deter-mined by multiple backward stepwise logistic regression analysis,in which only the independent variables were associated with LGA risk,with a P value<0.05.Finally,a risk prediction model was established and subsequently evaluated by using area under the receiver operating characteristic curve,calibration curve and decision curve analyses.RESULTS After using a multistep screening method,we establish a predictive model.Several risk factors for delivering an LGA infant were identified(P<0.01),including weight gain during pregnancy,parity,triglyceride-glucose index,free tetraiodothyronine level,abdominal circumference,alanine transaminase-aspartate aminotransferase ratio and weight at 24 gestational weeks.The nomogram’s prediction ability was supported by the area under the curve(0.703,0.709,and 0.699 for the training cohort,validation cohort,and test cohort,respectively).The calibration curves of the three cohorts displayed good agreement.The decision curve showed that the use of the 10%-60%threshold for identifying pregnant women with GDM who are at risk of delivering an LGA infant would result in a positive net benefit.CONCLUSION Our nomogram incorporated easily accessible risk factors,facilitating individualized prediction of pregnant women with GDM who are likely to deliver an LGA infant.

三硅氧烷羧酸钠表面活性剂的合成及性能

采用1,1,1,3,5,5,5-七甲基三硅氧烷(TSO)和丙烯酸乙酯(EA)为原料,通过硅氢加成和酯水解反应,合成一种新型三硅氧烷羧酸盐表面活性剂——七甲基三硅氧烷丙酸钠(TSO-SP)。通过傅里叶变换红外光谱仪(FT-IR)对目标产物进行结构表征,测试了产物的表面张力、泡沫性能、消抑泡性能、乳化性能、接触角和热稳定性能。结果表明:目标产物的临界胶束浓度(CMC)为7.94×10^(-3)mol/L,最低表面张力为26.18 mN/m;2 g/L的TSO-SP溶液的起泡量为2 mL,5 min内稳泡率为25%,TSO-SP对3种起泡液的消抑泡性能由大到小依次为十六烷基三甲基溴化铵(CTAB)、十二烷基苯磺酸钠(SDBS)、烷基糖苷-0814(APG-0814);TSO-SP对玉米油的乳化效果要远优于液体石蜡、二甲基硅油(350 Pa·s)和环己烷;1 g/L的TSO-SP溶液可以润湿固体石蜡表面,平衡接触角为72°;TSO-SP的热解温度为330℃,具有良好的热稳定性。

基于交互注意力和图卷积网络的方面级情感分析

方面级情感分析是细粒度情感分析中的一项关键任务,旨在预测一个句子中不同方面术语的情感倾向。针对目前结合图卷积网络的研究忽略方面术语本身的含义以及方面术语与上下文之间的交互的问题,文中提出了基于交互注意力和图卷积网络的模型(Interactive Attention Graph Convolution Network,IAGCN)。该模型首先结合BiLSTM和修正动态权重层对上下文进行建模,其次在句法依存树上使用图卷积网络对句法信息进行编码,然后利用交互注意力机制学习上下文和方面术语中的注意力,重构上下文和方面术语的表示,最后通过softmax层获取给定方面术语的情感极性。与基线模型相比,所提模型在5个数据集中的准确率和F1值分别提高了0.56%~1.75%和1.34%~4.04%。同时,将预训练模型BERT应用到此任务中,相比基于GloVe的IAGCN模型,其准确率和F1值分别提高了1.47%~3.95%和2.59%~7.55%,模型效果有了进一步的提升。

OsNPF3.1,a member of the NRT1/PTR family,increases nitrogen use efficiency and biomass production in rice

The overuse of nitrogen(N)fertilizer in fields has increased production costs and raised environmental concerns.Increasing the N use efficiency(NUE)of rice varieties is crucial for sustainable agriculture.Here we report the cloning and characterization of OsNPF3.1,a gene that controls rice NUE.An amino acid mutation in the OsNPF3.1 coding region caused different NUEs in wild and cultivated rice.OsNPF3.1,which is expressed mainly in the aerial parts of rice,also affects rice plant height,heading date,and thousand-grain weight.The OsNPF3.1 protein is located in the plasma membrane.When OsNPF3.1 was subjected to artificial selection,two naturally varying loci were associated with NUE,of which OsNPF3.1Chr6_8741040differed between indica and japonica rice.OsNPF3.1 can be used as a new target gene for breeding rice varieties with high NUE.

Transcriptome analysis reveals potential genes associated with plant height in rice

Plant height(PH)is a complex trait regulated by the environment and multiple genes.PH directly affects crop yield,harvest index,and lodging resistance.From plant dwarf mutants,many genes related to PH have been identified and described.Nonetheless,the molecular mechanism of height regulation in high-culm rice mutants has not been well studied.By using transcriptome and weighted gene co-expression network analysis(WGCNA),we identified the differentially expressed genes(DEGs)between high-culm rice mutants(MUT)and wild-type(WT)and explored the key pathways and potential candidate genes involved in PH regulation.Transcriptome analysis identified a total of 2,184 DEGs,of which 1,317 were identified at the jointing stage and 1,512 were identified at the heading stage.Kyoto Encyclopedia of Genes and Genomes enrichment showed that the enrichment pathways were mainly involved in plant hormone signal transduction,ABC transportation,and steroid hormone biosynthesis.Among these metabolic pathways,LOC_Os05g43910 and LOC_Os01g35030 were auxin(IAA)-related genes,up-regulated in MUT and LOC_Os02g08500(LEPTO1),LOC_Os11g04720,and LOC_Os12g04500 were cytokinin(CK)-related genes,downregulated in MUT.The WGCNA identified four modules(light cyan,dark grey,grey,and pale turquoise)closely related to PH,and seven key genes were screened from these modules,of which two were up-regulated cell wallrelated genes(LOC_Os01g26174(OsWAK5),LOC_Os06g05050)in MUT,and one gibberellic acid(GA)gene(LOC_Os06g37364,OsKO2)was also up-regulated.These genes might be closely related to PH regulation.These findings help us better understand the transcriptional regulation of rice plant growth and development and provide a theoretical basis for mapping and cloning the PH regulatory genes.

面向复杂场景的行人重识别综述

行人重识别(Person Re-Identification,简称Re-ID)旨在研究多个不相交摄像头间特定行人的匹配问题。文中首次以复杂场景中需要克服的挑战为行人重识别论文的分类依据,将2010-2021年期间发表的研究成果分为7类,即姿势问题、遮挡问题、照明问题、视角问题、背景问题、分辨率问题以及开放性问题,该分类方式有利于研究人员从实际需求出发,根据要解决的问题找到相应的解决方案。首先回顾行人重识别的研究背景、意义及研究现状,总结当前主流的行人重识别框架,统计了2013年以来发表在三大计算机视觉顶级会议CVPR,ICCV以及ECCV的论文情况和国家基金项目中Re-ID的相关项目情况;其次就复杂场景中面临的七大挑战,分别从问题成因和解决方案两方面对现有文献展开分析,归纳总结出处理各类挑战的主流方法;然后给出了行人重识别研究中泛化性较高的方法,并列举了当前行人重识别研究的难点;最后讨论了行人重识别未来的发展趋势。

跨领域文本的可迁移情绪分析方法

随着移动互联网的迅猛发展,社交网络平台充斥着大量带有情绪色彩的文本数据,对此类文本中的情绪进行分析研究不仅有助于了解网民的态度和情感,而且对科研机构和政府掌握社会的情绪变化及走向有着重要作用。传统的情感分析主要对情感倾向进行分析,无法精确、多维度地描述出文本的情绪,为了解决这个问题,文中对文本的情绪分析进行研究。首先针对不同领域文本数据集中情绪标签缺乏的问题,提出了一个基于深度学习的可迁移情绪分类的情感分析模型FMRo-BLA,该模型对通用领域文本进行预训练,然后通过基于参数的迁移学习、特征融合和FGM对抗学习,将预训练模型应用于特定领域的下游情感分析任务中,最后在微博的公开数据集上进行对比实验。结果表明,该方法相比于目前性能最好的RoBERTa预训练语言模型,在目标领域数据集上F1值有5.93%的提升,进一步加入迁移学习后F1值有12.38%的提升。

综合护理在急性胰腺炎患者中的应用效果研究

目的探究综合护理在急性胰腺炎患者中的护理效果。方法对本科2015年1月~2017年9月收治的96例急性胰腺炎患者进行研究,随机将其分为对照组(48例)和综合护理组(48例),在药物治疗基础上,对照组给予常规护理,综合护理组给予综合护理,比较2组在平均住院时间、脏器(肺脏、肾脏)衰竭发生率、死亡率的差异。结果经过本科综合护理措施,急性胰腺炎患者综合护理组较对照组平均住院时间缩短(P<0.05),脏器(肺脏、肾脏)衰竭发生率降低(P<0.05),治愈率提高(P<0.05)。结论对患者采取心理、身体机能、管道等方面综合护理措施,可缩短患者住院时间,降低患者死亡率,对急性胰腺炎患者的康复具有十分重要的作用。

DC铸造7475铝合金中含铬弥散相非均匀分布的机理

研究7475铝合金半连续铸锭中枝晶臂横截面上镁元素和铬元素的实际有效分配系数。同时,通过微观结构分析,揭示E(Al_(18)Mg_(3)Cr_(2))弥散相在半连续铸锭中非均匀分布的机理。结果表明,镁和铬的实际有效分配系数分别为0.650和1.392,这些系数表明合金枝晶臂横截面半径方向上镁和铬的非均匀分布。经470℃均匀化处理24 h后,镁元素扩散均匀,而铬元素几乎不扩散。镁和铬的浓度以及合金中的异质形核位置同时是形成E弥散相的决定因素。250℃热处理72 h为大量细小的E弥散相在随后高温热处理过程中合金枝晶臂横截面心部的形成提供异质形核位置。

Neurotoxicity and Biomarkers of Lead Exposure: a Review

Appropriate selection and measurement of lead biomarkers of exposure are critically important for health care management purposes,public health decision making,and primary prevention synthesis.Lead is one of the neurotoxicants that seems to be involved in the etiology of psychologies.Biomarkers are generally classified into three groups:biomarkers of exposure,effect,and susceptibility.The main body compartments that store lead are the blood,soft tissues,and bone;the half-life of lead in these tissues is measured in weeks for blood,months for soft tissues,and years for bone.Within the brain,lead-induced damage in the prefrontal cerebral cortex,hippocampus,and cerebellum can lead to a variety of neurological disorders,such as brain damage,mental retardation,behavioral problems,nerve damage,and possibly Alzheimer’s disease,Parkinson’s disease,and schizophrenia.This paper presents an overview of biomarkers of lead exposure and discusses the neurotoxic effects of lead with regard to children and adults.

Recovery of Collided RFID Tags With Frequency Drift on Physical Layer

In a passive ultra-high frequency(UHF)radio frequency identification(RFID)system,the recovery of collided tag signals on a physical layer can enhance identification efficiency.However,frequency drift is very common in UHF RFID systems,and will have an influence on the recovery on the physical layer.To address the problem of recovery with the frequency drift,this paper adopts a radial basis function(RBF)network to separate the collision signals,and decode the signals via FM0 to recovery collided RFID tags.Numerical results show that the method in this paper has better performance of symbol error rate(SER)and separation efficiency compared to conventional methods when frequency drift occurs.

Current progress in metabolomics of gestational diabetes mellitus

Gestational diabetes mellitus(GDM)is one of the most common metabolic disorders of pregnancy and can cause short-and long-term adverse effects in both pregnant women and their offspring.However,the etiology and pathogenesis of GDM are still unclear.As a metabolic disease,GDM is well suited to metabolomics study,which can monitor the changes in small molecular metabolites induced by maternal stimuli or perturbations in real time.The application of metabolomics in GDM can be used to discover diagnostic biomarkers,evaluate the prognosis of the disease,guide the application of diet or drugs,evaluate the curative effect,and explore the mechanism.This review provides comprehensive documentation of metabolomics research methods and techniques as well as the current progress in GDM research.We anticipate that the review will contribute to identifying gaps in the current knowledge or metabolomics technology,provide evidence-based information,and inform future research directions in GDM.

Factors influencing resilience in patients with burns during rehabilitation period

Purpose:To investigate the factors that influence levels of resilience in patients with burns during rehabilitation,and to provide theoretical guidance for psychological crisis prevention and intervention.Methods:A total of 129 patients with burns and undergoing rehabilitation were investigated using a demographic questionnaire,the Connor-Davidson Resilience Scale(CD-RISC),the Social Support Rating Scale(SSRS),and the Simplified Coping Styles Questionnaire.Results:The overall resilience score of burn patients was at a middle level.Multiple regression analysis showed that gender,marital status,occupation,educational level,burn severity,and a positive coping style were all significant factors influencing patient resilience.Conclusion:During psychological crisis intervention,medical staff should guide burn patients according to their individual coping styles.Such guidance would achieve a better effect,improve patient resilience,and promote positive psychological adaptation.

Efficacy of Yiqiyangxin Chinese medicine compound combined with cognitive therapy in the treatment of generalized anxiety disorders

Objective:To observe the clinical efficacy of Yiqiyangxin Chinese medicine compound combined with cognitive therapy on generalized anxiety disorders.Methods:A total of 202 generalized anxiety disorders patients were randomly allocated to a control condition(Paroxetine combined with cognitive therapy) or a treatment condition(Yiqiyangxin Chinese medicine compound combined with cognitive therapy).Subsequently,scores of Hamilton Anxiety Scale(HAMA), Zung Self-rating Anxiety Scale(SAS) and blood routine,urine routine,liver function,renal function,electrocardiogram were detected before treatment,3 months,6 months after treatment and 6 months after medicine withdrawal,respectively.Results:HAMA and SAS scores were significantly reduced in two groups(P【0.001) after treatment,but the differences in HAMA and SAS scores,apparent effect and effectiveness were not significant(P】0.05).HAMA and SAS scores were significantly increased in two groups(P【0.05) after medicine withdrawal,and there were significant differences in HAMA and SAS scores,recurrent disease and adverse reaction(P【0.001). The incidence of recurrent disease and adverse reaction in treatment group was low.Both two groups showed no apparent abnormality in blood routine,urine routine,liver and renal function, and electrocardiogram.Conclusions:Yiqiyangxin Chinese medicine compound combined with cognitive therapy can significantly reduce the recurrence after medicine withdrawal and is effective on generalized anxiety disorders.Furthermore,the incidence of adverse reactions is low. The treatment program is worthy clinic application in the further.

Placental Origins of Preeclampsia:Potential Therapeutic Targets

Preeclampsia(PE)remains a leading cause of maternal and perinatal morbidity and mortality in obstetrics worldwide.No effective treatments to reduce its incidence and severity in clinical practice are currently available.A variety of hypotheses have been generated aiming to explain the origins of PE,notably being the genetic predispositions and placental dysfunction.As regard to placental dysfunction,much progress has been made in basic research and several potential therapeutic targets have been identified.This review will discuss in detail the potential therapeutic targets in PE models including uteroplacental blood flow,oxidative stress,vasoactive factors and inflammation/immune response,and introduce the evolving technologies for placental research nowadays.

Capture-aware Bayesian RFID Tag Estimate for Large-scale Identification

Dynamic framed slotted Aloha algorithm is one of popular passive radio frequency identification(RFID) tag anticollision algorithms. In the algorithm, a frame length requires dynamical adjustment to achieve higher identification efficiency.Generally, the adjustment of the frame length is not only related to the number of tags, but also to the occurrence probability of capture effect. Existing algorithms could estimate both the number of tags and the probability of capture effect. Under large-scale RFID tag identification, however, the number of tags would be much larger than an initial frame length. In this scenario, the existing algorithm's estimation errors would substantially increase. In this paper, we propose a novel algorithm called capture-aware Bayesian estimate, which adopts Bayesian rules to accurately estimate the number and the probability simultaneously. From numerical results, the proposed algorithm adapts well to the large-scale RFID tag identification. It has lower estimation errors than the existing algorithms. Further,the identification efficiency from the proposed estimate is also higher than the existing algorithms.

Weight Based Channel Selection Towards 5G in the Unlicensed Spectrum

The growth of the wireless and mobile communication data traffic has brought severe challenges to the present telecommunication systems. To meet the ever-increasing mobile traffic demand in the next 5 th generation(5 G) communication systems, deploying 5 G in the unlicensed spectrum(5 G-U), has been regarded as a promising technology. The Third Generation Partnership Project(3 GPP) has specified the standardization of the Licensed Assisted Access(LAA) and its extension enhanced LAA(e LAA), to opportunistically transmit in the unlicensed spectrum. The LAA/e LAA systems share unlicensed spectrum resource with other networks, e.g., the Wi-Fi systems. In this article, we analyze the coexistence between the e LAA and the Wi-Fi systems in the unlicensed spectrum. We firstly establish the system model where the e LAA coexists with the Wi-Fi systems. Then, we theoretically derive and figure out the unfairness in the multi-channel occupancy rate between the e LAA and the Wi-Fi systems. After that, we propose a weight based channel selection method to improve the fairness of the coexistence. The numerical results demonstrate that by avoiding contentions and declining collisions, our method not only enhances the fairness, but also improves the overall unlicensed spectrum usage rate.

Validation of the Red Pericarp Gene from 419 Rice Landraces in Guangxi via Genome-wide Association Studies

Rice has different colors of pericarp, such as red, white and black. Red rice pericarp is rich in proanthocyanins, which have antioxidant properties and are beneficial to human health. In the present study, we analyzed the red-pericarp gene Rc of 419 rice landraces in Guangxi by genome-wide association study （GWAS）, and validated that the Rc gene regulated the red periearp trait in flee. By analyzing the genomie DNA of 97 red-pericarp flee eultivars, we identified two new alleles in C139 and C323. Then, the exons of Rcc＇9 and Rcc were sequenced with Sanger method, and the results demonstrated that the natural mutations within Re ene resulted in the two alleles Rcc and Rcc.