基于神经网络的语音识别研究

由于具有良好的抽象分类特性,神经网络现已应用于语音识别系统的研究和开发,并成为解决识别相关问题的有效工具.为解决一般语音识别系统准确率较低的问题,本文分别给出了由循环神经网络(RNN)和多层感知器(MLP)组成识别模块的两种语音识别系统,并对二者识别的准确性进行了比较.介绍了特征提取模块的主要工作步骤并讨论了组成识别模块的上述两种神经网络结构.其中,特征提取模块利用线性预测编码(LPC)倒谱编码器,把输入语音翻译成LPC倒谱空间中的曲线;而识别模块完成对某个特征空间曲线之间的联系和单词的识别.实验结果表明,MLP方法准确率高于RNN方法,而RNN方法准确率可达85%.     

基于双分支交互的实时语义分割算法

针对双分支实时语义分割算法存在双分支交互差、多尺度上下文信息提取不完善等问题,提出了基于双分支交互的实时语义分割网络(dual-branch interactive multi-scale fusion network for real-time semantic segmentation,DIMFNet)。算法以引导聚合双边语义分割网络(bilateral network with guided aggregation for real-time semantic segmentation, BiseNetV2)的双分支结构为基准进行改进,空间分支提取空间细节特征,上下文分支提取深层上下文特征。结合注意力思想提出注意力引导高级语义融合模块(attention guide high-level semantics fusion module,AGHSM)实现双分支的交互融合,以获得更好的空间特征表示;对金字塔池化模块进行改进,提出采用多层聚合金字塔池化模块(multi-layer aggregation pyramid pooling module,MAPPM）提取多尺度上下文特征,以...     

多尺度特征融合的图像语义分割

针对多尺度下的目标分割问题,提出了一种多尺度特征融合的图像语义分割方法.该方法改进了空洞空间金字塔池化(ASPP)模块,采用6种不同采样率空洞卷积并行架构增强了对输入图像信息的提取,该模块可以获取多尺度的卷积特征,从而进一步增强图像分割能力.通过在PASCAL VOC 2012数据集上进行训练和验证,达到了82.0%的平均像素准确率和71.9%的平均交并比.实验结果表明:通过改进空间维度上的ASPP模块,该方法可以达到良好的图像语义分割效果.     

改进的YOLOv3交通标志识别算法

针对复杂场景下交通标志检测存在精度低、检测速度慢等问题,提出一种基于YOLOv3改进的S-YOLO交通标志算法。首先,合并批归一化层到卷积层,以提升模型前向推理速度;其次,采用二分K-means聚类算法,确定适合交通标志的先验框;然后引入空间金字塔池化模块,提取特征图深度特征;最后引入CIoU回归损失函数,提升模型检测精度。实验结果表明,在重制的CTSDB交通标志数据集下,所提算法与YOLOv3相比,平均准确率和检测速度分别提升了4.26%和15.19%,同时相较YOLOv4以及其他算法对交通标志识别有更优的精度和速度,具有良好的鲁棒性,满足复杂场景高效实时检测。     

基于注意力机制与空间金字塔池化的行人属性识别

针对目前的行人属性识别方法存在鲁棒性差、特征表达能力不足和行人的细粒度特征识别精度不高的缺点,提出一种基于注意力机制与空间金字塔池化的行人属性识别方法,通过注意力机制强化不同维度的特征,提升行人整体特征表达;通过空间金字塔池化操作,实现任意大小图像的输入,更好地保留图像的特征信息。结果表明,与现有的其他方法相比,所提出的行人属性识别方法可以精确地识别行人多种属性,具有较高的行人细粒度特征识别精度。     

基于改进Bilinear-CNN的服装图像风格识别

为解决服装风格的细粒度识别问题,提出一种改进的Bilinear-CNN(convolutional neural network)模型。在VGG16特征提取网络的输出特征图中引入空间注意力机制,提升对图像中服装区域的关注程度。设计一种单特征通路的双线性池化方法,在减小参数量和计算量的同时,采用全局平均池化和全局最大池化两种操作充分挖掘细粒度特征。试验结果表明,改进的Bilinear-CNN模型在FashionStyle14数据集上达到76.4%的识别准确率。相比原始Bilinear-CNN模型,改进模型的识别准确率提升2个百分点,减少40%的参数量和52%的计算量,并对含扰动服装图像具有风格识别的稳健性。     

基于卷积神经网络的壁面清洗机器人障碍物检测识别算法

结合机器人的工作原理以及卷积神经网络(CNN)在图像分类中的应用,提出了一种基于卷积神经网络的壁面障碍物检测识别算法.首先,以壁面障碍物准确识别为目标,构建壁面障碍物图像库;然后,通过对VGG-16网络简化后进行优化,得到适合壁面障碍物准确识别的卷积神经网络模型.在此基础上,设计该网络由输入层、4层卷积层、2层池化层、1层全连接层以及输出层组成,进一步利用3×3卷积核对训练样本进行卷积操作,并将所获取的特征图以2×2领域进行池化操作.重复上述操作后,通过学习获取并确定网络模型参数,得到最佳网络模型.实验结果表明,障碍物的识别准确率可达99.0%,具有良好的识别能力.     

特征增强U形网络的图像语义分割

复杂场景语义分割任务是对场景图像逐像素进行分类并标记.图像中目标种类多,尺度多样的特点给分割任务增加了难度,提出了特征增强U形卷积神经网络(feature enhanced U shape networks,FEUNet)是一种改进的编码器加解码器的结构,编码阶段引入局部特征增强模块(local feature enhanced,LFE)提取局部感知特征来改善非显著目标的分割效果;考虑到神经网络深层和浅层之间特征表达的差异,在解码阶段利用全局池化方法(global pooling)设计全局特征增强模块(global feature enhanced,GFE),实现选择性地从深层特征图提取上下文信息作为对浅层特征图的指导,改善深层和浅层特征图的融合,保证同类像素预测的一致性.采用CamVid和Cityscapes数据集进行试验,模型mIOU测评值分别达到64.5％和73.2％,对比其他主流语义分割算法,该方法在分割性能和模型体积上具有一定竞争力.     

面向遥感图像的小样本目标检测改进算法研究

针对传统的遥感图像目标检测中面临的小样本以及目标样本分布不均衡等问题, 提出了一种基于改进的卷积神经网络(convolutional neural network, CNN)的遥感图像小样本目标检测算法. 首先, 该算法利用 $K$ 近邻($K$-nearest neighbor, kNN)回归分别对每个点和卷积层提取特征构建局部邻域; 同时, 通过最大池化聚合所有局部特征进行全局特征表示; 最后, 采用全连接层与缩放指数型线性单元(scaled expected linear unit, SELU)激活函数计算各类别对应的概率并分类. 实验结果表明, 该算法能够更有效地融合局部特征, 提高了遥感图像小样本目标识别与检测的精度, 同时保持信息的非局部扩散.     

基于深度卷积神经网络和支持向量机的手势识别算法

为了提高手势识别的准确率,提出一种基于深度卷积神经网络和支持向量机的手势识别算法;将包含手势的图像进行手掌轮廓分割及手指关节特征提取,经过去噪后获得准确的手势图像,然后通过卷积与池化获得手势特征样本,采用神经网络算法对输入特征样本进行训练,并对全连接层各节点的输出结果进行支持向量机多元分类,从而获得手势识别结果;在差异化设置条件下,通过对比手势识别的平均准确率和识别时间,可获得最优的卷积核尺寸及池化方法。仿真实验结果表明,相比其他3种识别算法,所提出的算法具有更优的识别准确率。     

Markedness and its Interpretation in Language Use

Language markedness is a common phenomenon in languages, and is reflected from hearing, vision and sense, i.e. the variation in the three aspects such as phonology, morphology and semantics. This paper focuses on the interpretation of markedness in language use following the three perspectives, i.e. pragmatic interpretation, psychological interpretation and cognitive interpretation, with an aim to define the function of markedness.     

Characterization of Ordovician carbonate reservoirs, southeastern Saskatchewan, Canada

The discovery of the prolific Ordovician Red River reservoirs in 1995 in southeastern Saskatchewan was the catalyst for extensive exploration activity which resulted in the discovery of more than 15 new Red River pools. The best yields of Red River production to date have been from dolomite reservoirs. Understanding the processes of dolomitization is, therefore, crucial for the prediction of the connectivity, spatial distribution and heterogeneity of dolomite reservoirs.The Red River reservoirs in the Midale area consist of 3～4 thin dolomitized zones, with a total thickness of about 20 m, which occur at the top of the Yeoman Formation. Two types of replacement dolomite were recognized in the Red River reservoir: dolomitized burrow infills and dolomitized host matrix. The spatial distribution of dolomite suggests that burrowing organisms played an important role in facilitating the fluid flow in the backfilled sediments. This resulted in penecontemporaneous dolomitization of burrow infills by normal seawater. The dolomite in the host matrix is interpreted as having occurred at shallow burial by evaporitic seawater during precipitation of Lake Almar anhydrite that immediately overlies the Yeoman Formation. However, the low δ18O values of dolomited burrow infills (-5.9‰～ -7.8‰, PDB) and matrix dolomites (-6.6‰～ -8.1‰, avg. -7.4‰ PDB) compared to the estimated values for the late Ordovician marine dolomite could be attributed to modification and alteration of dolomite at higher temperatures during deeper burial, which could also be responsible for its 87Sr/86Sr ratios (0.7084～0.7088) that are higher than suggested for the late Ordovician seawaters (0.7078～0.7080). The trace amounts of saddle dolomite cement in the Red River carbonates are probably related to "cannibalization" of earlier replacement dolomite during the chemical compaction.     

低渗透非均质砂岩油藏启动压力梯度研究

理论推导与室内实验相结合,建立了低渗透非均质砂岩油藏启动压力梯度确定方法。首先借助油藏流场与电场相似的原理,推导了非均质砂岩油藏启动压力梯度计算公式。其次基于稳定流实验方法,建立了非均质砂岩油藏启动压力梯度测试方法。结果表明:低渗透非均质砂岩油藏的启动压力梯度确定遵循两个等效原则。平面非均质油藏的启动压力梯度等于各级渗透率段的启动压力梯度关于长度的加权平均;纵向非均质油藏的启动压力梯度等于各渗透率层的启动压力梯度关于渗透率与渗流面积乘积的加权平均。研究成果可用于有效指导低渗透非均质砂岩油藏的合理井距确定,促进该类油藏的高效开发。     

On Tragic Consciousness In The Works By Ernest Hemingway

As an American modern novelist who were famous in the literary world, Hemingway was not a person who always followed the trend but a sharp observer. At the same time, he was a tragedy maestro, he paid great attention on existence, fate and end-result. The dramatis personae's tragedy of his works was an extreme limit by all means tragedy on the meaning of fearless challenge that failed. The beauty of tragedy was not produced on the destruction of life, but now this kind of value was in the impact activity. They performed for the reader about the tragedy on challenging for the limit and the death.     

A computer generator for randomly layered structures

AcomputergeneratorforrandomlylayeredstructuresYUJia shun1,2,HEZhen hua2(1.TheInstituteofGeologicalandNuclearSciences,NewZealand;2.StateKeyLaboratoryofOilandGasReservoirGeologyandExploitation,ChengduUniversityofTechnology,China)Abstract:Analgorithmisintrod…     

An Index mechanism for multi-scale view in spatial databases

There are numerous geometric objects stored in the spatial databases. An importance function in a spatial database is that users can browse the geometric objects as a map efficiently. Thus the spatial database should display the geometric objects users concern about swiftly onto the display window. This process includes two operations:retrieve data from database and then draw them onto screen. Accordingly, to improve the efficiency, we should try to reduce time of both retrieving object and displaying them. The former can be achieved with the aid of spatial index such as R-tree, the latter require to simplify the objects. Simplification means that objects are shown with sufficient but not with unnecessary detail which depend on the scale of browse. So the major problem is how to retrieve data at different detail level efficiently. This paper introduces the implementation of a multi-scale index in the spatial database SISP (Spatial Information Shared Platform) which is generalized from R-tree. The difference between the generalization and the R-tree lies on two facets: One is that every node and geometric object in the generalization is assigned with a importance value which denote the importance of them, and every vertex in the objects are assigned with a importance value,too. The importance value can be use to decide which data should be retrieve from disk in a query. The other difference is that geometric objects in the generalization are divided into one or more sub-blocks, and vertexes are total ordered by their importance value. With the help of the generalized R-tree, one can easily retrieve data at different detail levels.Some experiments are performed on real-life data to evaluate the performance of solutions that separately use normal spatial index and multi-scale spatial index. The results show that the solution using multi-scale index in SISP is satisfying.     

Achieving Unusual States of Matter Under High Pressure

正The periodicity of the elements and the non-reactivity of the inner-shell electrons are two related principles of chemistry,rooted in the atomic shell structure.Within compounds,Group I elements,for example,invariably assume the+1 oxidation state,and their chemical properties differ completely from those of the p-block elements.These general rules govern our understanding of chemical structures and reactions.Using first principles calcula-     

Exchange-Correlation and Electronic Excitation Energies from Pairing Matrix Fluctuations

We have developed an adiabatic connection to formulate the ground-state exchange-correlation energy in terms of pairing matrix linear fluctuations.This formulation of the exchange-correlation energy opens a new channel for density functional approximations based on the many-body perturbation theory.We illustrate the potential of such approaches with an approximation based on the particle-particle Random Phase Approximation(pp-RPA).This re-     

Density Functional Theory for the Response of Periodic Systems to Electric Fields Based on the Vector Potential Approach

正The electronic and nuclear(structural/vibrational)response of 1D-3D nanoscale systems to electric fields gives rise to a host of optical,mechanical,spectral,etc.properties that are of high theoretical and applied interest.Due to the computational difficulty of treating such large systems it is convenient to model them as infinite and periodic(at least,in first approximation).The fundamental theoretical/computational problem in doing so is that     

On the Thermodynamic Limit for Responses to Static Electromagnetic Fields

For molecular systems,the quantum-mechanical treatment of their responses to static electromagnetic fields usually employs a scalar-potential treatment of the electric field and a vector-potential treatment of the magnetic field.Although the potential for each field separately is associated with the choice of an(unphysical)origin,the precise choice of the origin for the electrostatic field has little consequences for the results.This is different for the