基于Ghost-SE-Res2Net的多模型融合语音唤醒词检测方法

语音唤醒词检测(WWD)是语音交互中的关键技术,选择合适大小的检测窗对WWD性能的影响很大。提出一种新的多模型融合方法,通过融合小检测窗和大检测窗的检测结果来提高WWD性能。多模型融合方法包含两个分类模型,分别使用小检测窗和大检测窗,均基于轻量化的挤压与激励残差网络(SE-Res2Net)模块,即GhostSE-Res2Net,SE-Res2Net结构的多尺度机制可显著提升WWD的能力。在Ghost-SE-Res2Net中,首先使用Ghost卷积替换SE-Res2Net中的普通卷积以降低模型参数量,然后使用注意力池化层替换SE-Res2Net中的全局平均池化层进一步提升WWD能力。在实际检测时融合连续3个小检测窗模型的检测结果的最大值和1个大检测窗模型的检测结果,来判断唤醒词是否被触发。在训练时引入困难样本挖掘算法,选择性地学习较难检测的唤醒词信息以提高分类模型的检测性能。在包含2个唤醒词的Mobvoi数据集上评估系统性能,实验结果表明,在每小时0.5次错误唤醒的情况下,该系统在2个唤醒词上的错误拒绝率分别为0.46%和0.43%,实现了与先进基线相似的性能,并且系统参数量比基线少31%。

基于Ghost-YOLOv5s的SAR图像舰船目标检测

基于星载合成孔径雷达(synthetic aperture radar,SAR)图像的舰船目标检测中,为了平衡模型大小与检测精度,提出了一种基于Ghost卷积的SAR图像舰船目标检测方法Ghost-YOLOv5s。在YOLOv5s的颈部引入Ghost卷积,以减少模型参数和压缩模型体积;将高效的通道注意力机制(efficient channel attention,ECA)融入到颈部的C3模块里,以突出重要特征,从而保持较高的检测性能;使用SIoU损失函数替换原来的CIoU损失函数,以减少预测框和真实框之间的偏差,提高检测算法精度。实验结果表明,在SSDD遥感数据集上,改进模型与YOLOv5s相比,模型参数量减少了6.28%,模型体积减小了6.21%,检测精度达到了98.21%,实现了模型大小与检测精度的平衡。

On the Stringy Ghosts Which We Call the Missing Dark Energy of the Cosmos

Dark energy is explained using familiar notions and concepts used in quantum field theory, string theory and the exact mathematical theory of spacetime. The main result of the present work is first a new mathematical definition of pre-quantum spacetime (QST) as a multiset made of infinitely many empty Cantor sets connected to pre-quantum wave empty set (QW) and the pre-quantum particle (QP) zero set via the cobordism equation ∂(QW) = (QP)U(QST). Second, and in turn, this new path of reasoning is used to validate the quantum splitting of Einstein’s E = mc2 into the sum of the ordinary energy E = mc2/22 of the quantum particle and the dark energy E = mc2(21/22) of the quantum wave, used predominantly to explain the observed accelerated expansion of the universe.

Study on Genetic Inactivation Bacterial Ghosts of Pasteurella multocida based PhiX174 Gene E Lysis Cassette Mediated

[Objective] To study on genetic inactivation bacterial ghosts of Pasteurella multocida based PhiX174 gene E lysis cassette mediated. [ Method ] Recombinant pPBA1100-e was constructed by which the gene E of bacteriophage Phix174 and temperature sensitivity expressing control system hybridized with plasmid pPBA1100 by genetic engineering method. Recombinant was transformed to Pasteurella multocida and lysis gene E expressed by temperature induction. Recombinant was detected by restriction endonuclease. Cell morphology of bacterial ghost of Pasteurella mul- tocida was observed by scanning electron microscopy and inactivation ratio was estimated by CFU analysis. I Result~ The results indicated that the recombination plasmid presented three bands by restriction endonuclease and agarose electrophoresis and that molecular weight of every band ac- corded with theoretical value. The result of SEM observing showed that recombination plasmid expressed successfully in P. multocida and produced bacterial ghost. The result of CFU detecting demonstrated that inactivation ratio of P. multocida reached 99 per cent. ~Conclusion~ This study pro- vided technical bases for the preparation of antigen vaccine of natural bacterial outer membrane protein.

基于Ghost模块的农资图像文本检测算法及其应用

针对农资图像中文本的检测速度慢并且缺乏移动端的应用等问题,基于农资图像数据集,提出了一种基于Ghost模块的农资图像文本检测算法,该算法对DB网络进行改进,使用MobileNetv2网络来提取基础特征,引入多尺度特征融合模块来获得多层之间的特征融合,并采用可微分二值化后处理算法预测文本,使其能够快速地检测农资图像中的文本。该算法在农资图像数据集上的准确率基本达到了主流算法的标准,检测速度达18.6 img/s,参数量为2.99 M,具备轻量级的特征,将此算法部署到移动端设备上并成功运行。

To Appreciate and Comprehend the Ghosts Elements in A Christmas Carol

This paper explores the message hidden in the ghost characters in Dickens'A Christmas Carol. There are an immense number of books about ghost both in Chinese and British literature. Ghost is an imaginary afterlife of human being; similarly novel is also a kind of fiction based on the real life. Therefore,when novels are used to shape the ghost images,there will be a powerful and unstrained space produced by the double-tier imagination. The ghost novels can also serve as a mirror of human society and speak of some problems in human life. So a better understanding of the ghost characters in literature can give us a deep insight into the society at the time when the story took place,as well as providing us a good chance of literature appreciation.Considering the role that the ghosts play in this novel,it should be called memory figures,developing Scrooge from a character wishing to shut out his past to one living in the past,the present,and the future. The first indication of his character can be derived from the remark that he takes "his melancholy dinner in his usual melancholy tavern" and then returns to his "gloomy suite of rooms ". The adjectives tell us that Scrooge should have a talent for remem?brance or rather a good memory. His humoral constitution predisposes him for it. The three ghosts,whose visitation Marley announced,each take Scrooge to places,to towns and buildings,the most traditional type of image in artificial memory. It is typical of these visits that Scrooge,like Vergil in Dante's Divine Comedy,becomes an invisible visitor,who enters the world of his past evoked by the memorial places. This immediate contact appears as a precondition for the reformative influence of the past on Scrooge. If we look back to an earlier Christmas story,which also preaches the change of heart,Dickens' new and specific frame of reference shows up even more clearly. (Cerny,257)

GHOSTS AND GORE A closer look at depiction of women in Thai horror movies

The 1999 Thai horror movie Nang Nak was the turning point for the“new wave”of Thai movies,and it became a model for the horror genre in Thailand.Even today,vengeance from female ghosts remains a central theme of Thai horror movies.

Ghosts in the shell:identification of microglia in the human central nervous system by P2Y12 receptor

Microglia are the tissue resident macrophages of the brain and represent the sole immune population located in the parenchyma of the central nervous system （CNS）. These cells are hidden be-tween neurons, astrocytes as well as oligodendrocytes and account for only 5-10% of CNS cells. Even though microglia were already identified in 1913 by the Spanish neuroanatomist Ramon y Cajal and further seminally investigated by his student Pio del Rio Hortega,

Joint Authentication Public Network Cryptographic Key Distribution Protocol Based on Single Exposure Compressive Ghost Imaging

In the existing ghost-imaging-based cryptographic key distribution(GCKD)protocols,the cryptographic keys need to be encoded by using many modulated patterns,which undoubtedly incurs long measurement time and huge memory consumption.Given this,based on snapshot compressive ghost imaging,a public network cryptographic key distribution protocol is proposed,where the cryptographic keys and joint authentication information are encrypted into several color block diagrams to guarantee security.It transforms the previous single-pixel sequential multiple measurements into multi-pixel single exposure measurements,significantly reducing sampling time and memory storage.Both simulation and experimental results demonstrate the feasibility of this protocol and its ability to detect illegal attacks.Therefore,it takes GCKD a big step closer to practical applications.

High-quality ghost imaging based on undersampled natural-order Hadamard source

Improving the speed of ghost imaging is one of the main ways to leverage its advantages in sensitivity and imperfect spectral regions for practical applications.Because of the proportional relationship between image resolution and measurement time,when the image pixels are large,the measurement time increases,making it difficult to achieve real-time imaging.Therefore,a high-quality ghost imaging method based on undersampled natural-order Hadamard is proposed.This method uses the characteristics of the Hadamard matrix under undersampling conditions where image information can be fully obtained but overlaps,as well as deep learning to extract aliasing information from the overlapping results to obtain the true original image information.We conducted numerical simulations and experimental tests on binary and grayscale objects under undersampling conditions to demonstrate the effectiveness and scalability of this method.This method can significantly reduce the number of measurements required to obtain high-quality image information and advance application promotion.

High-visibility ghost imaging with phase-controlled discrete classical light sources

We take phase modulation to create discrete phase-controlled sources and realize the super-bunching effect by a phasecorrelated method. From theoretical and numerical simulations, we find the space translation invariance of the bunching effect is a key point for the ghost imaging realization. Experimentally, we create the orderly phase-correlated discrete sources which can realize high-visibility second-order ghost imaging than the result with chaotic sources. Moreover, some factors affecting the visibility of ghost image are discussed in detail.

Computational temporal ghost imaging based on complementary modulation

We report an experimental demonstration of temporal ghost imaging in which a digital micromirror device(DMD)and+1/-1 binary modulation have been combined to give an accurate reconstruction of a nonperiodic time object.Compared to the 0/1 modulation,the reconstruction signal can be improved greatly by+1/-1 binary modulation even with half of the measurements.Experimental results show that 0/1 binary temporal objects up to 4 kHz and sinusoidal time objects up to 1 kHz can be reconstructed by this method.The influences of modulation speed and array detector gray levels are also discussed.

基于Ghost卷积的高级别浆液性卵巢癌复发预测方法

高级别浆液性卵巢癌是一种恶性肿瘤疾病,进行术前复发预测能帮助临床医生为患者提供个性化治疗方案,降低病人的死亡率。因该疾病的医学数据较少且难以获取,导致其深度学习模型难以得到充分的训练,复发预测准确率有待提高。针对此问题,本文设计了一种改进的低参数残差网络TGE-ResNet34,以ResNet34为主干网络,将传统卷积模块用Ghost卷积代替,完成病灶区特征的提取,降低模型的参数量,在2个Ghost卷积之间融入ECA(Efficient Channel Attention)注意力机制,抑制无用特征提取的干扰,最后通过5折交叉验证模型,避免数据随机划分的偶然性。实验结果表明,改进设计的TGE-ResNet34网络准确率为96.01%,相比原基线网络准确率提高4.52个百分点,参数量减少15.98 M。

基于相位变换和GhostNet-门控循环单元的自动调制识别方法

针对信号调制方式低信噪比条件下识别准确率不高的问题,提出了一种由相位变换、GhostNet、压缩与激励网络(Squeeze and Excitation Network,SENet)、门控循环单元(Gated Recurrent Unit,GRU)和深度神经网络组成的模型,用于自动调制识别接收信号。首先,采用基准数据集RML2016.10a和RML2016.10b同相正交数据作为模型输入;其次,构建识别模型,其中,相位变换用于降低相位偏移对调制识别的影响,GhostNet和GRU分别用于提取调制信号的空间特征和时间特征,SENet用于对特征图权重进行调整;而后,通过深度神经网络进行分类;最后,对所提模型进行了训练及测试。实验结果表明:与现有模型CGDNet、CLDNN、IC-AMCNet、MCLDNN和LSTM相比,所提出模型显著降低了参数量,有效提升了低信噪比条件下的识别准确率,平均识别准确率分别达到62.30%和64.45%。

Windows 7万能Ghost启动盘仿真软件的设计与实现

传统的操作系统安装方法不仅费时费力,而且对一般用户来说也颇有难度,不易掌握,所以制作容易操作的系统安装光盘很有必要。该文利用Ghost、Easy Sysprep、万能驱动助理等工具软件,制作了Windows7万能Ghost启动盘。并以Authorware多媒体软件为开发工具,设计了交互式仿真软件,用以模拟万能系统盘的制作过程,这对一般用户学习和掌握相关技术具有一定参考价值。

自适应笛卡尔网格Ghost Cell方法研究

在笛卡尔网格中,利用Ghost Cell方法处理浸入边界,模拟二维无粘可压缩流。针对静止物体,比较了各种不同的Ghost Cell边界条件下的熵误差,总压误差,以及阻力系数。此外为提高激波分辨率,将Ghost Cell方法与基于叉树结构的自适应笛卡尔网格算法相结合,数值结果显示本文的方法是切实可行的。

三维流体界面不稳定性的Ghost方法

Euler方法对大多数气体流动处理得很好 ,但在临近材料界面的区域出现非物理振荡 .Lagarange方法对多材料界面适用性很好 ,但它在处理大变形和大多数气体流动的涡结构却有自身的困难 .理想的稳定性强的方法是结合两种方法的长处 .利用LevelSet方法来确定界面位置 ,利用Ghost方法来确定临近界面的网格点的状态量 .采用单侧界面的熵插值技巧来捕捉适宜的界面边界条件 。

克隆软件Ghost在图书馆电子阅览室中的应用

本文介绍了克隆软件Ghost的几种工作模式及在实际工作中的应用 。

Symantec Ghost技术在网络管理中的应用解析

在全球的占有率达70％的Symantec Ghost 7.5提供对诸如PC布局、网络克隆、改变配置选项、用户移植、备份和磁盘映像恢复等多种任务的远程控制管理,从而简化并提高了对网络化PC系统管理的效率。本文着重分析了Symantec Ghost系统中的网络克隆功能,同时也对此系统其它方面作了简要描述。