Adapter Based on Pre-Trained Language Models for Classification of Medical Text

We present an approach to classify medical text at a sentence level automatically.Given the inherent complexity of medical text classification,we employ adapters based on pre-trained language models to extract information from medical text,facilitating more accurate classification while minimizing the number of trainable parameters.Extensive experiments conducted on various datasets demonstrate the effectiveness of our approach.

A Spectral Convolutional Neural Network Model Based on Adaptive Fick’s Law for Hyperspectral Image Classification

Hyperspectral image classification stands as a pivotal task within the field of remote sensing,yet achieving highprecision classification remains a significant challenge.In response to this challenge,a Spectral Convolutional Neural Network model based on Adaptive Fick’s Law Algorithm(AFLA-SCNN)is proposed.The Adaptive Fick’s Law Algorithm(AFLA)constitutes a novel metaheuristic algorithm introduced herein,encompassing three new strategies:Adaptive weight factor,Gaussian mutation,and probability update policy.With adaptive weight factor,the algorithmcan adjust theweights according to the change in the number of iterations to improve the performance of the algorithm.Gaussianmutation helps the algorithm avoid falling into local optimal solutions and improves the searchability of the algorithm.The probability update strategy helps to improve the exploitability and adaptability of the algorithm.Within the AFLA-SCNN model,AFLA is employed to optimize two hyperparameters in the SCNN model,namely,“numEpochs”and“miniBatchSize”,to attain their optimal values.AFLA’s performance is initially validated across 28 functions in 10D,30D,and 50D for CEC2013 and 29 functions in 10D,30D,and 50D for CEC2017.Experimental results indicate AFLA’s marked performance superiority over nine other prominent optimization algorithms.Subsequently,the AFLA-SCNN model was compared with the Spectral Convolutional Neural Network model based on Fick’s Law Algorithm(FLA-SCNN),Spectral Convolutional Neural Network model based on Harris Hawks Optimization(HHO-SCNN),Spectral Convolutional Neural Network model based onDifferential Evolution(DE-SCNN),SpectralConvolutionalNeuralNetwork(SCNN)model,and SupportVector Machines(SVM)model using the Indian Pines dataset and PaviaUniversity dataset.The experimental results show that the AFLA-SCNN model outperforms other models in terms of Accuracy,Precision,Recall,and F1-score on Indian Pines and Pavia University.Among them,the Accuracy of the AFLA-SCNN model on Indian Pines reached 99.875%,and the Accuracy on PaviaUniversity reached 98.022%.In conclusion,our proposed AFLA-SCNN model is deemed to significantly enhance the precision of hyperspectral image classification.

ADASYN与类别逆比例加权法在阿尔茨海默病不平衡数据中的应用

目的利用自适应合成抽样(adaptive synthetic sampling,ADASYN)与类别逆比例加权法处理类别不平衡数据,结合分类器构建模型对阿尔茨海默病(alzheimer′s disease,AD)患者疾病进程进行分类预测。方法数据源自阿尔茨海默病神经影像学计划(Alzheimer′s disease neuroimaging initiative,ADNI),经随机森林填补缺失值,弹性网络筛选特征子集后,利用ADASYN与类别逆比例加权法处理类别不平衡数据。分别结合随机森林(random forest,RF)、支持向量机(support vector machine,SVM)构建四种模型:ADASYN-RF、ADASYN-SVM、加权随机森林(weighted random forest,WRF)、加权支持向量机(weighted support vector machine,WSVM),与RF、SVM比较分类性能。模型评价指标为宏观平均精确率(macro-average of precision,macro-P)、宏观平均召回率(macro-average of recall,macro-R)、宏观平均F1值(macro-average of F1-score,macro-F1)、准确率(accuracy,ACC)、Kappa值和AUC(area under the ROC curve)。结果ADASYN-RF的分类性能最优(Kappa值为0.938,AUC为0.980),ADASYN-SVM次之。利用ADASYN-RF预测得到的重要分类特征分别为CDRSB、LDELTOTAL、MMSE,在临床上均可得到证实。结论ADASYN与类别逆比例加权法都能辅助提升分类器性能,但ADASYN算法更优。

融合专利SAO特征的早期专利价值评估方法

[研究目的]在知识产权强国战略背景下,存在着专利数量与质量不匹配、技术转化率较低的困境,评估早期专利价值对高价值专利培育工程和专利技术转化具有重要意义。[研究方法]提出一种融合专利SAO网络特征和文本特征的早期专利价值评估方法。首先,从专利摘要和标题中抽取主体-行为-客体(SAO)结构。其次,用社会网络分析法挖掘专利SAO网络的核心组件,得到专利SAO网络中心性特征。然后,将SAO结构序列文本输入Doc2Vec模型,得到专利SAO文本特征。最后,融合专利基础特征和SAO特征,构建基于AutoGluon自动机器学习框架的早期专利价值评估模型。[研究结论]以人工智能产业专利为例进行实证研究,实验结果表明引入专利SAO特征后,模型的准确率、召回率和F1值相比基准模型均提高了2%~5%,证明了专利SAO特征在识别早期高价值专利中的有效性。

基于S-CAD法的县域新型城镇化发展规划方案评估研究——以陕西省佳县为例

规划方案作为相关利益主体价值观的直接体现,应当在规划评估中予以重视。基于城乡规划的公共政策属性,采用S-CAD政策评估方法,以陕西省佳县的《新型城镇化与城乡统筹规划》文本为例,选取规划编制与实施主体——政府作为主导视角,对规划文本的内容进行解构,在此基础上进行规划立场、目标、手段与预期结果的一致性、充要性与依赖性分析。研究发现:一、该规划文本自身的逻辑性较强,各个环节存在基本的对应关系;二、规划文本当中对于产业发展以及由此带来的人口问题、基础设施与公共服务设施问题的解决方案描述较多,对于生态环境保护、交通环境改善以及人口流动的实质性解决方案相对较少,逻辑上存在一定的漏洞,需进一步完善逻辑链条。基于此,提出建立城乡规划全过程评估体系及重视多元利益主体协调关系的建议,以期为相关规划的编制与评估工作提供参考。

基于知识融合和聚类引导的语言模型用于MOFs合成信息分类

金属有机框架(MOFs)的合成实验步骤通常集中存储在科学文献某一段落内,从文献中提取实验步骤对推动新型金属有机框架的开发具有重要意义。现有研究存在两个问题:第一,将整篇文献视为普通文本,按句/段直接切分处理,忽略了上下文中隐藏的高级知识;第二,模型未深入挖掘数据内部的隐藏模式。针对上述问题,提出一个基于知识融合策略的高质量知识补充任务。利用科学文献编辑风格和结构化Web数据的微妙之处,将上下文关键知识汇集到段落中,进而优化其文本表征;然后提出一种基于聚类引导的自适应分类算法,采用聚类算法将文本表征划分为多个簇,同时训练不同的分类模型来适应不同的簇,从而有效地减少数据重叠的影响,提高模型召回率。实验结果表明,所提方法的性能相比主流基线模型有较大提升。

Effect of bowel rehabilitative therapy on structural adaptation of remnant small intestine: animal experiment

AIM: To investigate the individual and the combined effects of glutamine, dietary fiber, and growth hormone on the structural adaptation of the remnant small bowel. METHODS: Forty-two adult male Sprague-Dawley rats underwent 85% mid-small bowel resection and received total parenteral nutrition (TPN) support during the first three postoperational days.From the 4th postoperational day, animals were randomly assigned to receive 7 different treatments for 8 days: TPNcon group, receiving TPN and enteral 20 g x L(-1) glycine perfusion; TPN+Gln group, receiving TPN and enteral 20 g x L(-1) glutamine perfusion; ENcon group, receiving enteral nutrition (EN) fortified with 20 g x L(-1) glycine; EN+Gln group, enteral nutrition fortified with 20 g x L(-1) glutamine; EN+Fib group, enteral nutrition and 2 g x d(-1) oral soybean fiber; EN+GH group, enteral nutrition and subcutaneous growth hormone (GH) (0.3 IU) injection twice daily; and ENint group, glutamine-enriched EN, oral soybean fiber, and subcutaneous GH injection. RESULTS: Enteral glutamine perfusion during TPN increased the small intestinal villus height (jejunal villus height 250 microm +/- 29 microm in TPNcon vs 330 microm +/- 54 microm in TPN+Gln, ileal villus height 260 microm +/- 28 microm in TPNcon vs 330 microm +/- 22 microm in TPN+Gln, P【0.05) and mucosa thickness (jejunal mucosa thickness 360 microm +/- 32 microm in TPNcon vs 460 microm +/- 65 microm in TPN+Gln, ileal mucosa thickness 400 microm +/- 25 microm in TPNcon vs 490 microm +/- 11 microm in TPN+Gln,P【 0.05) in comparison with the TPNcon group. Either fiber supplementation or GH administration improved body mass gain (end body weight 270 g +/- 3.6g in EN+Fib, 265.7 g +/- 3.3 g in EN+GH, vs 257 g +/- 3.3 g in ENcon, P【 0.05), elevated plasma insulin-like growth factor (IGF-I) level (880 microg x L(-1). 52 microg x L-(-1) in EN+Fib,1200 microg x L(-1). 96 microg x L-(-1) in EN +/- GH, vs 620 microg x L(-1).43 microg x L-(-1) in ENcon, P【 0.05), and increased the villus height (jejunum 560 microm +/- 44 microm in EN +/- Fib, 530 microm +/- 30 microm in EN +/- GH, vs 450 microm +/- 44 microm in ENcon, ileum 400 microm +/- 30 microm in EN+Fib, P【0.05) and the mucosa thickness (jejunum 740 microm +/- 66 microm in EN +/- Fib, 705 microm +/- 27 microm in EN +/- GH, vs 608 microm +/- 58 microm in ENcon, ileum 570 microm +/- 27 microm in EN +/- Fib, 560 microm +/- 56 microm in remnant jejunum and ileum. Glutamine-enriched EN produced little effect in body mass, plasma IGF-I level, and remnant small bowel mucosal structure. The ENint group had greater body mass (280 g +/- 2.2g), plasma IGF-I level (1450 microg x L(-1). 137 microg x L-(-1)), and villus height (jejunum 620 microm +/- 56 microm, ileum 450 microm +/- 31 microm) and mucosal thickness (jejunum 800 microm +/- 52 microm, ileum 633 microm +/- 33 microm) than those in ENcon, EN+Gln (jejunum villus height and mucosa thickness 450 microm +/- 47 microm and 610 +/- 63 microm, ileum villus height and mucosa thickness 330 microm +/- 39 microm and 500 microm +/- 52 microm), EN+GH groups (P【0.05), and than those in EN+Fib group although no statistical significance was attained. CONCLUSION: Both dietary fiber and GH when used separately can enhance the postresectional small bowel structural adaptation. Simultaneous use of these two gut-trophic factors can produce synergistic effects on small bowel structural adaptation. Enteral glutamine perfusion is beneficial in preserving small bowel mucosal structure during TPN, but has little beneficial effect during EN.

Research on the effects of family care degree on stigma and psychosocial adaptation among the patients with Parkinson’s disease

Objective:This study aimed to investigate the status quo of family care degree,stigma and psychosocial adaptation of the patients with Parkinson’s disease(PD),and discuss the effects of family care degree on stigma and psychosocial adaptation among the patients with PD.Methods:This study was conducted between September 2018 and April 2019 on patients with PD who were admitted to a Ter tiary Grade A hospital in Dalian city.A total of 246 PD patients were investigated by convenience sampling method using self-designed general information questionnaire,along with family care index(APGAR),24-item Stigma Scale for Chronic Illness(SSCI),and Psychosocial Adjustment Scale(PAS)for PD.And patients were divided into two groups according to the level of family care degree and compared with stigma and psychosocial adaptation.Results:The total score of family care for patients with PD was 6.97±2.46,family function was good in 132 cases(53.7%),moderate barrier in 92 cases(37.4%),and severe barrier in 22 cases(8.9%).The scores of stigma and psychosocial adaptability and their dimensions in the group with high degree family care were higher than the low degree group.Conclusions:The overall level of family care for PD patients was poor.Improving family care degree can help patients with PD to reduce stigma and improve psychosocial adaptation.

改进YOLOX-S的偏光片表面缺陷检测算法

偏光片是液晶显示器的重要组成部分,其表面缺陷不仅会降低液晶显示器的显示质量,甚至可能造成整个液晶面板的报废。针对偏光片表面缺陷存在尺度差异大、形状变化多样的问题,提出一种改进YOLOX-S的偏光片表面缺陷检测算法。提出自适应平衡特征金字塔(ABFP)模块充分融合主干网提取的多级特征,并通过单个卷积增加检测分支,进一步增强模型的多尺度检测能力。在ABFP中引入注意力模块CBAM关注重要特征。采用CIo U损失函数的同时使用Mish激活函数替代Si LU激活函数。实验结果表明,改进的算法在偏光片表面缺陷数据集上的m AP_(50)和m AP_(50:95)分别达到92.97%和55.16%,相比YOLOX-S(FPN)提升了1.86和1.34个百分点,每秒检测帧数(FPS)达到50,基本满足工业实时检测的需求。

HAL1 mediate salt adaptation in Arabidopsis thaliana

The yeast HAL1 gene was introduced into Arabidopsis thaliana by Agrobacterium tumefaciens-mediated transformation with vacuum infiltration under the control of CaMV 35S promoter. Thirty-three individual kanamycin resistant plants were obtained from 75,000 seeds. Southern blotting analysis indicated that HAL1 gene had been integrated into all of the transgenic plants’ genomes. The copy number of HAL1 gene in transgenic plants was mostly 1 to 3 by Southern analysis. Phenotypes of transgenic plants have no differences with wild type plants. several samples of transformants were self-pollinated, and progenies from transformed and non-transformed plants (controls) were evaluated for salt tolerance and gene expression. Measurement of concentrations of intracellular K+ and Na+ showed that transgenic lines were able to retain less Na+ than that of the control under salt stress. Results from different tests indicated the expression of HAL1 gene promotes a higher level of salt tolerance in vivo in the transgenic Arabidopsis plants.

基于T-S MPC的车辆自适应巡航控制策略研究

针对于汽车自适应巡航系统面对复杂工况时自适应性不强的问题,提出了一种基于T-S模糊变权重模型预测控制(Takagi-Sugeno fuzzy model predictive control,T-S MPC)的自适应巡航分层控制策略。首先上层控制器基于安全距离模型将自适应巡航系统划分为定速巡航模式、多目标优化控制的跟随模式和紧急制动模式;下层控制器基于车辆逆动力学模型,将上层控制器输出的期望加速度转变为节气门开度或制动压力;其次考虑到权重系数对控制精度的影响,建立基于T-S模糊控制的变权重模型预测控制器;最后搭建Carsim Simulink联合仿真平台,验证控制策略的准确性、适应性和跟踪响应速度。结果表明,在定速巡航工况时,T-S MPC控制方法跟踪响应时间为1.54 s,较PID和传统MPC控制跟踪响应快;在跟随和混合工况时,T-S MPC控制方法均方根误差为分别为0.3073、2.775,均低于PID、PID+LQR和传统MPC控制的均方根误差、并且自适应性好,有效提高了跟车性能与安全性。

一种S波段超宽带能量选择表面设计

设计了一种工作在S波段的能量选择表面,可实现超宽带自适应强电磁防护。该结构由两层金属周期结构组成,顶层为两个对称分布的金属条和一个金属片,金属条与金属片间加载两个PIN二极管;底层为十字架结构。当入射电磁波场强低于阈值时,能量选择表面工作在透波状态,电磁波可以传播;当入射电磁波场强超过阈值时,金属条和金属片之间产生的感应电压使得PIN二极管导通,此时能量选择表面进入防护状态,电磁波被屏蔽。通过对能量选择表面在PIN二极管导通和截止状态下的表面电流和电场分布以及等效电路模型进行分析,解释了该结构的工作原理。采用PCB制作工艺加工了实物样板并对弱场入射下的插入损耗以及强场入射下的防护效能进行测试。实验和仿真结果匹配性良好,表明该能量选择表面在透波状态下的工作中心频率为2.7 GHz,插入损耗小于1 dB的工作频带为2.2~3.5 GHz;在防护状态下,工作频带的防护效能大于10 dB,达到了超宽带的要求。

Adapted Metrics for a Modified Coulomb/Newton’s Potential

Modified Theories of Gravity include spin dependence in General Relativity, to account for additional sources of gravity instead of dark matter/energy approach. The spin-spin interaction is already included in the effective nuclear force potential, and theoretical considerations and experimental evidence hint to the hypothesis that Gravity originates from such an interaction, under an averaging process over spin directions. This invites to continue the line of theory initiated by Einstein and Cartan, based on tetrads and spin effects modeled by connections with torsion. As a first step in this direction, the article considers a new modified Coulomb/Newton Law accounting for the spin-spin interaction. The physical potential is geometrized through specific affine connections and specific semi-Riemannian metrics, canonically associated to it, acting on a manifold or at the level of its tangent bundle. Freely falling particles in these “toy Universes” are determined, showing an interesting behavior and unexpected patterns.

基于自适应S面算法的小型ROV艏向和深度运动控制研究

带缆遥控水下机器人(ROV)系统艏向与深度运动具有强非线性和运行环境不确定性等特点。针对ROV的运动特点,构建运动控制模型并进行化简,通过构建自适应S面控制法对艏向与深度进行控制,并利用Lyapunov的稳定性判据证明该控制法的稳定性,通过仿真验证了该控制法具有良好的稳定性和控制品质。同时,与S面控制法相比,自适应S面控制具有更好的动态性能与静态性能,调节速度快、稳定性强,能够准确控制ROV运动。

基于S变换相关度的配电网单相接地自适应有源消弧法

提出了基于S变换相关度的配电网单向接地自适应有源消弧法,有效消除了配电网单相接地故障电弧,提升了电力系统稳定性。采集配电网线路零序电流时序数据,经S变换获取零序电流时频信息矩阵后,计算并获取各线路间时频信息矩阵相关度矩阵,输入卷积神经网络模型,输出单相接地故障选线结果;在故障线路采用基于模糊准PR与PI控制的双闭环自适应控制方法,向配电网故障线路中性点注入零序电流,有效控制注入电流相位以及幅值,迫使配电网故障线路所在相的电压值为0,完成自适应有源消弧。实验结果表明:该方法可有效识别配电网单相接地故障线路,实现配电网单相接地自适应有源消弧,且可尽快完成单相接地自适应有源消弧工作,保证配电网安全稳定运行。

Cross-Cultural Adjustment and Coping:U.S.Students in Taiwan Residents Higher Education

The aim of this study was to investigate the adjustment problems of students from the United States enrolled in universities in the East,specifically in Taiwan,their problems related to cultural adaptation,and the process of adjustment to student life in Taiwan.Under investigation were cultural adjustment and coping skills as these students transitioned from West to East.Qualitative data were collected from interviews with participants and faculty members as well as participant observations.Results indicated that U.S.students found their own ways to acclimate to their new academic setting as well as to social relations,cross-cultural issues,and the linguistic environment in Taiwan to achieve effective adaptation.They made changes in themselves to cope with all situations they encountered.This study provides suggestions for international students abroad in Taiwan,for the Taiwan Residents government,and for universities or colleges in terms of what they should offer to current and future international students.

基于B/S架构的空调智能监控系统设计与实现

为解决空调系统设备种类众多、协议不统一的问题,针对空调系统的监控需求,设计了基于分布式监控与集中管理模式、B/S架构的空调智能监控系统。首先,在设备监控层利用多线程技术实现设备实时数据的采集处理,采用多协议转换技术进行统一协议转换、封装,将多种协议数据上传存储到数据库服务器。然后,通过模糊自适应PID控制器对监测数据进行轮询计算,调节变频器实时控制空调系统温度。最后,通过Web服务端口访问数据库服务器,实现系统集成与数据交换,建立集中控制、管理、运维的集成平台,实现系统设备的统一监控和调度,系统的统一部署和高可用性,完成空调监控系统的界面可视化并提升空调系统智能化水平。实践表明,该系统运行稳定,能实时监控空调运行状况,且具有较高的扩展性与管理效率。

The Impact of Climate Change on Rainfall Patterns in Ghana:A Zoning Adaptation Strategy through Developing Agroforestry

As a developing country in Africa,the effect of climate change is one of the sufferings of Ghana.The effect is much felt in rainfall variability because of the country over reliance on rainfall for agriculture.While various researches have studied the impact of climate change in Ghana,few among them have extended to its impact on rainfall pattern across all the ecological zones in the country.The trends in the rainfall from seven selected meteorological stations across all the ecological zones were analyzed using data from the NASA satellite.The Mann-Kendall and Sen’s Slope Test were used for the analysis.The study found decreasing trend in most of the monthly and yearly rainfall pattern across the ecological zones,and suggested cashew agroforestry as a zoning adaptation strategy.

制冷站系统负荷和能效比改进T-S模糊模型构建

制冷站具有非线性、强耦合等特点,导致传统机理的建模面临困难。为实现系统节能优化控制,并改善模型在线修正和移植性能,提出一种基于改进Takagi-Sugeno(T-S)模糊模型的制冷站负荷和能效比动态建模方法,前件结构辨识设计改进天牛须搜索算法,以改善模糊C-均值聚类方法对初值敏感和易陷入局部最优的问题;为实现非线性模型辨识,并降低现场测试数据噪声的影响,设计自适应扩展卡尔曼滤波算法实现模型后件参数辨识和在线修正。实验结果表明,所建立的负荷和能效比预测模型在广州某建筑上运行时,相对误差分别为0.63%和1.49%;使用广州市另一座建筑的数据进行模型可移植性验证,经过500步在线训练,新模型成功收敛,证明所构建模型具备良好的可移植性和适应性。

基于稀疏自适应S变换的特种设备故障振动信号检测研究

鉴于特种设备的特殊性,其故障的检测尤为重要,而检测特种设备振动信号是发现故障的一种重要手段,基于此种原因,本文提出基于稀疏自适应S变换特种设备故障振动信号检测方法。从特种设备的振动信号时频特征出发,通过稀疏自适应S变换提取特种设备振动信号时频特征图;构建深度卷积神经网络模型,将经过稀疏自适应S变换提取的时频特征图作为网络模型输入样本,经深度学习后,完成特种设备故障振动信号检测,获取设备故障诊断结果。实验结果表明:该方法提取到振动信号特征较好,可清晰表达故障频率,特征表达能力强;可明确检测出特种设备故障发生时间以及故障原因,且检测准确率高。