Neural Network and Method of Training
    3.
    发明申请
    Neural Network and Method of Training 审中-公开
    神经网络和训练方法

    公开(公告)号:US20060112028A1

    公开(公告)日:2006-05-25

    申请号:US10711191

    申请日:2004-11-24

    IPC分类号: G06N3/02

    CPC分类号: G06N3/08

    摘要: Methods of training neural networks (100, 600) that include one or more inputs (102-108) and a sequence of processing nodes (110, 112, 114, 116) in which each processing node may be coupled to one or more processing nodes that are closer to an output node are provided. The methods include establishing an objective function that preferably includes a term related to differences between actual and expected output for training data, and a term related to the number of weights of significant magnitude. Training involves optimizing the objective function in terms of weights that characterize directed edges of the neural network. The objective function is optimized using algorithms that employ derivatives of the objective function. Algorithms for accurately and efficiently estimating derivatives of the summed input going into output processing nodes of the neural network with respect to the weights of the neural network are provided.

    摘要翻译: 训练包括一个或多个输入(102-108)和一系列处理节点(110,112,114,116)的神经网络(100,600)的方法,其中每个处理节点可以耦合到一个或多个处理节点 提供更接近于输出节点。 所述方法包括建立目标函数,其优选地包括与训练数据的实际输出和期望输出之间的差异相关的术语,以及与重要量级的权重数相关的术语。 训练涉及根据表征神经网络的有向边缘的权重来优化目标函数。 使用采用目标函数的导数的算法优化目标函数。 提供了用于精确有效地估计相对于神经网络的权重进入神经网络的输出处理节点的求和输入的导数的算法。

    Balancing workloads in an electronics assembly factory
    4.
    发明授权
    Balancing workloads in an electronics assembly factory 失效
    平衡电子组装工厂的工作量

    公开(公告)号:US06829514B2

    公开(公告)日:2004-12-07

    申请号:US10346693

    申请日:2003-01-17

    IPC分类号: G06F1900

    CPC分类号: G06Q10/06 H05K13/085

    摘要: A method of optimization of a product assembly line utilizes a mixed-integer linear programming (MILP) formulation. Line information, such as product data of products to be assembled by the line, parts data of the parts to be used to assemble the products, line data descriptive of the product assembly line and machine data of the machines of the line, is compiled and used to formulate a balancing strategy of the product assembly line (210, 240). The balancing strategy is formulated by the selective manipulation of one or more optimization variables to make the MILP closely representative of the actual manufacturing environment (240, 250, 270). Solving the MILP generates MILP output information from which output reports and metrics may be generated (270, 280).

    摘要翻译: 产品装配线的优化方法采用混合整数线性规划(MILP)公式。 编制线路信息,例如要组装的产品的产品数据,用于组装产品的部件的零件数据,描述产品装配线的线数据和线的机器的机器数据, 用于制定产品装配线(210,240)的平衡策略。 平衡策略是通过选择性地操纵一个或多个优化变量来制定的,以使MILP与实际的制造环境(240,250,270)密切相关。 解决MILP生成MILP输出信息,从中可以生成输出报告和度量(270,280)。

    Smart call list
    5.
    发明申请
    Smart call list 审中-公开
    智能通话列表

    公开(公告)号:US20070135110A1

    公开(公告)日:2007-06-14

    申请号:US11296953

    申请日:2005-12-08

    IPC分类号: H04M3/00

    摘要: A smart call list for a wireless mobile device (12) can include a storage medium (14 or 304) containing a call list having a plurality of contact entries (19) and a processor (302) coupled to the storage medium. The processor can monitor (902) the use of the call list according to a plurality of criteria to intelligently improve the chances of successfully contacting the entries on the call list. The processor can rank (903) the plurality of criteria and assign (906) relative weights to the plurality of criteria to facilitate a selection and order of the contact entries to call (904). The processor can further automatically select (908) a next best number to call if an attempt to connect with a current number is unsuccessful. In one arrangement, the smart call list can use a phone tree (30) in an attempt to contact members of the call list.

    摘要翻译: 无线移动设备(12)的智能呼叫列表可以包括包含具有耦合到存储介质的多个联系人条目(19)和处理器(302)的呼叫列表的存储介质(14或304)。 处理器可以根据多个标准监视(902)呼叫列表的使用,以智能地提高成功接触呼叫列表上的条目的机会。 处理器可对(903)多个标准进行排序(903),并将(906)相对权重分配给多个标准,以促进对呼叫的联系人条目的选择和顺序(904)。 如果连接当前号码的尝试不成功,处理器可以进一步自动选择(908)下一个最佳号码来呼叫。 在一种布置中,智能呼叫列表可以使用电话树(30)来尝试联系呼叫列表的成员。

    METHOD AND APPARATUS FOR ENHANCING GRAPHICAL USER INTERFACE APPLICATIONS
    7.
    发明申请
    METHOD AND APPARATUS FOR ENHANCING GRAPHICAL USER INTERFACE APPLICATIONS 审中-公开
    用于增强图形用户界面应用的方法和装置

    公开(公告)号:US20080010534A1

    公开(公告)日:2008-01-10

    申请号:US11382132

    申请日:2006-05-08

    IPC分类号: G06F11/00

    CPC分类号: G06F9/451 G06F9/44505

    摘要: A system and method (500) are disclosed for a method and apparatus for manipulating GUIs. A system that incorporates teachings of the present disclosure may include, for example, a computing device (100) having a controller (108) that manages a display (110). The controller can be programmed to alter (518) a graphical user interface (GUI) (200-400) presented on the display according to a behavior model that predicts (508-510) an expectation of an end user from observations of the end user's interactions with the GUI. Additional embodiments are disclosed.

    摘要翻译: 公开了一种用于操纵GUI的方法和装置的系统和方法(500)。 结合本公开的教导的系统可以包括例如具有管理显示器(110)的控制器(108)的计算设备(100)。 控制器可以被编程为根据行为模型改变(518)呈现在显示器上的图形用户界面(GUI)(200-400),行为模型预测(508-510)终端用户对最终用户的期望 与GUI的交互。 公开了另外的实施例。

    Method and apparatus for managing data
    8.
    发明申请
    Method and apparatus for managing data 审中-公开
    用于管理数据的方法和装置

    公开(公告)号:US20050033719A1

    公开(公告)日:2005-02-10

    申请号:US10634080

    申请日:2003-08-04

    CPC分类号: G06F16/25

    摘要: A data management system (100) including a knowledge container creator module (118) operative to create at least a first data descriptor item (112) and at least a second data descriptor item (114) based upon a raw data item (110). The raw data item (110) is capable of containing data representing raw data that is in one of a plurality of different formats. The knowledge container creator module (118) also operative to link the raw data item (110) to at the least a first data descriptor item (112) and to link the raw data item (110) to the at least a second data descriptor item (114).

    摘要翻译: 一种数据管理系统(100),包括知识容器创建器模块(118),用于基于原始数据项(110)创建至少第一数据描述符项(112)和至少第二数据描述符项(114)。 原始数据项(110)能够包含表示以多种不同格式之一的原始数据的数据。 知识容器创建器模块(118)还可操作以将原始数据项(110)链接到至少第一数据描述符项(112),并将原始数据项(110)链接到至少第二数据描述符项 (114)。