中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

104 records (took 0.043 seconds)

    Discriminative training in machine learning
    1.
    发明授权
    Discriminative training in machine learning 有权
    机器学习中的歧视性培训

    公开(公告)号：US08027938B1

    公开(公告)日：2011-09-27

    申请号：US12055967

    申请日：2008-03-26

    Applicant: Peng Xu ,  Ioannis Tsochandaridis

    Inventor： Peng Xu ,  Ioannis Tsochandaridis

    IPC: G06F15/18

    CPC classification number: G06N99/005

    Abstract: Systems, methods, and apparatuses including computer program products for machine learning are provided. A method is provided that includes distributing a parameterized model to each worker of a hierarchy of workers, the parameterized model including a plurality of feature functions and corresponding model parameters, processing a portion of training data at each worker of the plurality of workers according to the parameterized model to calculate updates to model parameters, for each worker at a lowest level of the hierarchy of workers, sending the calculated updates to a next higher level worker, for each other worker in the hierarchy of workers, combining updates of the respective worker with updates received from one or more lower level workers, collecting all updates from the workers at a master to generate real updates to the model parameters, and generating an updated model using the real updates to the model parameters.

    Abstract translation: 提供了包括用于机器学习的计算机程序产品的系统，方法和装置。 提供了一种方法，其包括将参数化模型分发给工作人员层级的每个工人，所述参数化模型包括多个特征功能和对应的模型参数，根据所述参数化模型处理所述多个工人的每个工人的一部分训练数据 参数化模型来计算模型参数的更新，对于工人层次结构的最低级别的每个工作者，将计算的更新发送给下一个更高级别的工作者，对于工人层次结构中的每个其他工作者，将相应工作者的更新与 从一个或多个较低级别的工作人员收到的更新，收集主人员的所有更新，以生成对模型参数的真实更新，以及使用模型参数的真实更新生成更新的模型。

    System and method for detecting non-contact repellency of a compound candidate from Drosophila
    2.
    发明授权
    System and method for detecting non-contact repellency of a compound candidate from Drosophila 有权

    公开(公告)号：US10948398B1

    公开(公告)日：2021-03-16

    申请号：US15986097

    申请日：2018-05-22

    Applicant: Peng Xu

    Inventor： Peng Xu

    IPC: A01M1/20 ,  A01M1/00 ,  G01N17/00 ,  A01M1/10

    Abstract: An assay system for a repellent includes transparent tubing having end chambers; a central testing area positioned between the end chambers; small openings between the central testing area and end chambers; absorbent material positioned adjacent to the openings and being soaked in either a control substance or a repellent; and fruit flies are introduced into the central testing area and allowed to migrate through the openings and into either end chamber.

    Training acoustic models using distributed computing techniques
    6.
    发明授权
    Training acoustic models using distributed computing techniques 有权
    使用分布式计算技术训练声学模型

    公开(公告)号：US08959014B2

    公开(公告)日：2015-02-17

    申请号：US13539225

    申请日：2012-06-29

    Applicant: Peng Xu ,  Fernando Pereira ,  Ciprian I. Chelba

    Inventor： Peng Xu ,  Fernando Pereira ,  Ciprian I. Chelba

    IPC: G06F17/21 ,  G06F17/20 ,  G10L15/08 ,  G10L15/14 ,  G10L15/187 ,  G10L15/04 ,  G10L15/34

    CPC classification number: G10L15/187 ,  G10L15/063 ,  G10L15/14 ,  G10L15/34 ,  G10L2015/0631

    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training acoustic models. Speech data and data identifying a transcription for the speech data are received. A phonetic representation for the transcription is accessed. Training sequences are identified for a particular phone in the phonetic representation. Each of the training sequences includes a different set of contextual phones surrounding the particular phone. A partitioning key is identified based on a sequence of phones that occurs in each of the training sequences. A processing module to which the identified partitioning key is assigned is selected. Data identifying the training sequences and a portion of the speech data are transmitted to the selected processing module.

    Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的用于训练声学模型的计算机程序。 接收用于识别语音数据的转录的语音数据和数据。 访问转录的语音表示。 在语音表示中为特定电话识别训练序列。 每个训练序列包括围绕特定电话的不同的上下文电话组。 基于在每个训练序列中出现的电话序列来识别分区密钥。 选择分配了所识别的分区键的处理模块。 识别训练序列和语音数据的一部分的数据被发送到所选择的处理模块。

    DELAY CIRCUIT AND ASSOCIATED METHOD
    8.
    发明申请
    DELAY CIRCUIT AND ASSOCIATED METHOD 有权
    延时电路及相关方法

    公开(公告)号：US20130257502A1

    公开(公告)日：2013-10-03

    申请号：US13431595

    申请日：2012-03-27

    Applicant: Yan Dong ,  Peng Xu

    Inventor： Yan Dong ,  Peng Xu

    IPC: H03H11/26

    CPC classification number: H03K5/133 ,  H02M3/1588 ,  H03K17/284 ,  Y02B70/1466

    Abstract: The embodiments of the present invention disclose a delay circuit. The delay circuit comprises an inverter, a load capacitor, and a first voltage clamping module, wherein the first voltage clamping module generates a voltage drop configured to prolong the propagation delay time of the delay circuit as the power supply voltage decreases. The power supply dependent delay circuit may have a much larger propagation delay time at low power supply voltage than it at high power supply voltage at the rising-edge or falling-edge of an input signal.

    Abstract translation: 本发明的实施例公开了延迟电路。 所述延迟电路包括反相器，负载电容器和第一电压钳位模块，其中所述第一电压钳位模块产生电压降，所述电压降配置为当所述电源电压降低时延长所述延迟电路的传播延迟时间。 电源相关延迟电路在低电源电压下的输入信号上升沿或下降沿处的高电源电压可能具有大得多的传播延迟时间。

    Metal Layout of an Integrated Power Transistor and the Method Thereof
    9.
    发明申请
    Metal Layout of an Integrated Power Transistor and the Method Thereof 审中-公开
    集成功率晶体管的金属布局及其方法

    公开(公告)号：US20130168869A1

    公开(公告)日：2013-07-04

    申请号：US13339005

    申请日：2011-12-28

    Applicant: Peng Xu

    Inventor： Peng Xu

    IPC: H01L23/522 ,  H01L21/768

    CPC classification number: H01L23/5226 ,  H01L23/4824 ,  H01L23/5283 ,  H01L2924/0002 ,  H01L2924/00

    Abstract: The present disclosure discloses a metal layout of an integrated power transistor. The metal layout comprises a 1st metal layer, a 2nd metal layer, and a 3rd metal layer. The metal layout couples the 1st metal layer to the 2nd metal layer through vias, and couples the 2nd metal layer to the 3rd metal layer through super vias. By such interconnection, the metallization resistance is highly reduced by using thick 2nd and 3rd metal layers.

    Abstract translation: 本公开公开了一种集成功率晶体管的金属布局。 金属布局包括第一金属层，第二金属层和第三金属层。 金属布局通过通孔将第一金属层耦合到第二金属层，并且通过超级通孔将第二金属层耦合到第三金属层。 通过这样的互连，通过使用厚的第2和第3金属层来高度降低金属化电阻。

    Method For Seismic Interpretation Using Seismic Texture Attributes
    10.
    发明申请
    Method For Seismic Interpretation Using Seismic Texture Attributes 审中-公开
    使用地震纹理属性的地震解释方法

    公开(公告)号：US20120090834A1

    公开(公告)日：2012-04-19

    申请号：US13378018

    申请日：2010-05-04

    Applicant: Matthias Imhof ,  Peng Xu

    Inventor： Matthias Imhof ,  Peng Xu

    IPC: G01V1/28 ,  E21B43/00 ,  E21B47/00

    CPC classification number: G06K9/00536 ,  G01V1/302 ,  G01V1/307 ,  G01V1/32 ,  G01V2210/43 ,  G01V2210/44 ,  G01V2210/63

    Abstract: Method for generating a new family of seismic attributes sensitive to seismic texture that can be used for classification and grouping of seismic data into seismically similar regions. A 2D or 3D data analysis window size is selected (23), and for each of multiple positions (25) of the analysis window in the seismic data volume, the data within the window are transformed to a wavenumber domain spectrum (26). At least one attribute of the seismic data is then defined based on one or more spectral properties, and the attribute is computed (28) for each window, generating a multidimensional spectral attribute data volume (29). The attribute data volume can be used for inferring hydrocarbon potential, preferably after classifying the data volume cells based on the computed attribute, partitioning the cells into regions based on the classification, and prioritizing of the regions within a classification.

    Abstract translation: 用于生成对地震纹理敏感的新的地震属性系列的方法，可用于将地震数据分类和分组到地震相似区域。 选择2D或3D数据分析窗口大小（23），并且对于地震数据卷中分析窗口的多个位置（25）中的每一个，将窗口内的数据转换为波数域频谱（26）。 然后基于一个或多个光谱属性来定义地震数据的至少一个属性，并且为每个窗口计算（28）属性，生成多维光谱属性数据卷（29）。 属性数据量可以用于推断碳氢化合物的潜力，最好是根据计算出的属性对数据体积单元进行分类，根据分类将单元细分成多个区域，以及分类中区域的优先级。

Patent Agency Ranking