Overexpression of an F-box Protein Gene Reduces Abiotic Stress Tolerance and Promotes Root Growth in Rice

As one of the largest gene families, F-box domain proteins have important roles in regulating various developmental processes and stress responses. In this study, we have investigated a rice F-box domain gene, MAIF1. The MAIF1protein is mainly localized in the plasma membrane and nucleus. MAIF1 expression is induced rapidly and strongly by abscisic acid (ABA) and abiotic stresses. MAIM expression is also induced in root tips by sucrose, independent of its hydrolytic hexose products, glucose and fructose, and the plant hormones auxin and cytokinin. Overexpression of MAIF1 reduces rice ABA sensitivity and abiotic stress tolerance and promotes rice root growth. These results suggest that MAIF1 is involved in multiple signaling pathways in regulating root growth. Growth restraint in plants is an acclimatization strategy against abiotic stress. Our results also suggest that MAIF1 plays the negative role in response to abiotic stress possibly by regulating root growth.     

7075合金挤压带板裂纹的探讨

7075合金挤压带板表面出现大小不一及分布不均的裂纹,低倍观察与分析发现其不仅大小不一,且方向都是垂直于挤压方向．认为它是挤压裂纹。本文主要针对制品发生的横向挤压裂纹进行分析并判定其产生的原阒及预防措施。     

基于深度残差收缩网络的HEp-2图像识别

人上皮细胞（HEp-2）检测抗核抗体是诊断自身免疫性疾病的常用方法,HEp-2细胞图像识别对许多自身免疫性疾病的诊疗具有重要意义。针对目前主要采用手工评估方法造成效率低效、劳动强度高等问题,提出一种基于深度残差收缩网络的HEp-2细胞图像分类模型。该模型在深度残差网络基础上进行改进,残差学习模块使用恒等映射方法可以训练更深层次的网络。在每个残差学习模块内部嵌入一个软阈值非线性变换子网络,软阈值用以消除数据中的噪声和冗余信息,这些阈值通过子网络自动学习。实验表明,该方法具有良好的性能,优于其他深度神经网络方法。     

分层特征融合注意力网络图像超分辨率重建

目的 深层卷积神经网络在单幅图像超分辨率任务中取得了巨大成功。从3个卷积层的超分辨率重建卷积神经网络（super-resolution convolutional neural network,SRCNN）到超过300层的残差注意力网络（residual channel attention network,RCAN）,网络的深度和整体性能有了显著提高。然而,尽管深层网络方法提高了重建图像的质量,但因计算量大、实时性差等问题并不适合真实场景。针对该问题,本文提出轻量级的层次特征融合空间注意力网络来快速重建图像的高频细节。方法 网络由浅层特征提取层、分层特征融合层、上采样层和重建层组成。浅层特征提取层使用1个卷积层提取浅层特征,并对特征通道进行扩充;分层特征融合层由局部特征融合和全局特征融合组成,整个网络包含9个残差注意力块（residual attention block,RAB）,每3个构成一个残差注意力组,分别在组内和组间进行局部特征融合和全局特征融合。在每个残差注意力块内部,首先使用卷积层提取特征,再使用空间注意力模块对特征图的不同空间位置分配不同的权重,提高高频区域特征的注意力,以快速恢复高频细节信息;上采样层使用亚像素卷积对特征图进行上采样,将特征图放大到目标图像的尺寸;重建层使用1个卷积层进行重建,得到重建后的高分辨率图像。结果 在Set5、Set14、BSD（Berkeley segmentation dataset）100、Urban100和Manga109测试数据集上进行测试。当放大因子为4时,峰值信噪比分别为31.98 dB、28.40 dB、27.45 dB、25.77 dB和29.37 dB。本文算法比其他同等规模的网络在测试结果上有明显提升。结论 本文提出的多层特征融合注意力网络,通过结合空间注意力模块和分层特征融合结构的优势,可以快速恢复图像的高频细节并且具有较小的计算复杂度。     

基于近邻关系聚合的人脸聚类方法

人脸聚类是人脸标注、人脸识别等工作的预处理过程。其主要对人脸图像进行分组,用来为人脸识别模型提供高质量的标注信息,从而有效降低人工标注的成本。人脸聚类的关键在于如何学习大规模人脸数据中整体及局部的结构关系,并把其迁移至待标注数据集。针对这一问题,本文提出一种基于近邻关系聚合的人脸聚类方法(Nearest Neighborhood Aggregation Clustering, NNAC)。该方法把局部结构的学习建模为一个近邻关系预测问题,通过堆叠多个改进的基于残差-全连接模块(Residual Fully-Connected Block, ResFCB)以提取多尺度的邻接关系特征。实验结果表明,相比主流人脸聚类方法,该方法在基准数据集上能够有效提升聚类精度。     

MANET 中基于链路稳定性路由的跨层传输控制协议

在MANET中,通信节点的移动会造成端到端通信路由的时常中断.传统TCP协议只有拥塞控制机制,对由于节点移动造成的数据包传输丢失和超时也作为拥塞处理,使得端到端传输性能低下.为解决这一问题,采用跨层设计思想,将传输控制与链路稳定性路由结合,提出了一种基于链路生存时间概率的传输控制协议(transmissioncontrol protocol based on probability of link residual lifetime,简称TCP-PLRT).该协议通过在路由稳定时跨层收集路由层链路生存时间概率信息来实现对端到端连接稳定性的认知,并针对路由不稳定、路由中断、路由恢复,分别制定了路由切换、数据存储转发、ACK再确认三大机制.这使得TCP-PLRT协议具有对由移动造成的路由中断进行提前预判和有效处理的能力.仿真结果表明,TCP-PLRT协议能够极大地减小由节点移动带来的端到端传输性能的下降,减少分组重传,提高端到端的吞吐量.     

一种基于Taylor和Kalman的室内协同定位方法

结合Chan算法、Taylor算法及Kalman算法三种TDOA算法的优点,提出一种能应用于室内实时定位的协同方法。首先基于Chan与Taylor的协同定位方法估算位置信息,并通过对估计结果的残差设置阈值来鉴别NLOS,从而抛弃受到NLOS污染严重的测量数据。其次,再对符合条件的测量数据,利用Kalman方法计算定位结果,与Taylor方法的定位结果通过设置判别条件进行比较,以此进一步抑制NLOS干扰。对符合判别条件的定位结果,进行残差加权及移动平均加权处理,从而完成最终定位结果的更新。最后,利用室内实时定位实验,证明该方法能有效过滤受到NLOS污染严重的测距数据,提高定位精度,并且具有良好的稳定性。     

基于显著性和最大熵的输送带撕裂检测

针对带式输送机在运行过程中输送带易撕裂的背景,根据人眼视觉系统特点和信息论相关理论,提出了一种基于视觉显著性和一维最大熵的输送带撕裂故障检测算法.首先在频域提取图像的显著性区域,然后计算显著图的最大信息熵,进而分割图像.实验结果表明,该方法具有较强的稳定性和适应性.     

ILB®, a Low Molecular Weight Dextran Sulphate,Restores Glutamate Homeostasis,Amino Acid Metabolism and Neurocognitive Functions in a Rat Model of Severe Traumatic Brain Injury

In a previous study, we found that administration of ILB^®, a new low molecular weight dextran sulphate, significantly improved mitochondrial functions and energy metabolism, as well as decreased oxidative/nitrosative stress, of brain tissue of rats exposed to severe traumatic brain injury (sTBI), induced by the closed-head weight-drop model of diffused TBI. Using aliquots of deproteinized brain tissue of the same animals of this former study, we here determined the concentrations of 24 amino acids of control rats, untreated sTBI rats (sacrificed at 2 and 7 days post-injury) and sTBI rats receiving a subcutaneous ILB^® administration (at the dose levels of 1, 5 and 15 mg/kg b.w.) 30 min post-impact (sacrificed at 2 and 7 days post-injury). Additionally, in a different set of experiments, new groups of control rats, untreated sTBI rats and ILB^®-treated rats (administered 30 min after sTBI at the dose levels of 1 or 5 mg/kg b.w.) were studied for their neurocognitive functions (anxiety, locomotor capacities, short- and long-term memory) at 7 days after the induction of sTBI. Compared to untreated sTBI animals, ILB^® significantly decreased whole brain glutamate (normalizing the glutamate/glutamine ratio), glycine, serine and γ-aminobutyric acid. Furthermore, ILB^® administration restored arginine metabolism (preventing nitrosative stress), levels of amino acids involved in methylation reactions (methionine, L-cystathionine, S-adenosylhomocysteine), and N-acetylaspartate homeostasis. The macroscopic evidences of the beneficial effects on brain metabolism induced by ILB^® were the relevant improvement in neurocognitive functions of the group of animals treated with ILB^® 5 mg/kg b.w., compared to the marked cognitive decline measured in untreated sTBI animals. These results demonstrate that ILB^® administration 30 min after sTBI prevents glutamate excitotoxicity and normalizes levels of amino acids involved in crucial brain metabolic functions. The ameliorations of amino acid metabolism, mitochondrial functions and energy metabolism in ILB^®-treated rats exposed to sTBI produced significant improvement in neurocognitive functions, reinforcing the concept that ILB^® is a new effective therapeutic tool for the treatment of sTBI, worth being tested in the clinical setting.