公开(公告)号：US08301638B2

公开(公告)日：2012-10-30

申请号：US12238012

申请日：2008-09-25

申请人： Ning-Yi Xu ,  Feng-Hsiung Hsu ,  Rui Gao ,  Xiong-Fei Cai ,  Junyan Chen

发明人： Ning-Yi Xu ,  Feng-Hsiung Hsu ,  Rui Gao ,  Xiong-Fei Cai ,  Junyan Chen

IPC分类号： G06F7/00 ,  G06F17/30 ,  G06F15/18

CPC分类号： G06F17/30864 ,  G06N99/005

摘要： A method using a RankBoost-based algorithm to automatically select features for further ranking model training is provided. The method reiteratively applies a set of ranking candidates to a training data set comprising a plurality of ranking objects having a known pairwise ranking order. Each round of iteration applies a weight distribution of ranking object pairs, yields a ranking result by each ranking candidate, identifies a favored ranking candidate for the round based on the ranking results, and updates the weight distribution to be used in next iteration round by increasing weights of ranking object pairs that are poorly ranked by the favored ranking candidate. The method then infers a target feature set from the favored ranking candidates identified in the iterations.

摘要翻译： 提供了一种使用基于RankBoost的算法自动选择特征进行进一步排名模型训练的方法。 该方法重复地将一组排名候选应用于包括具有已知成对排序顺序的多个排序对象的训练数据集。 每轮迭代应用排序对象对的权重分布，由每个排名候选者产生排名结果，根据排名结果识别轮次的优选排名候选者，并通过增加下一次迭代更新要使用的权重分布 排名对象对的权重由受欢迎的排名候选人排名较差。 该方法然后从迭代中识别的优选排名候选推断目标特征集。

公开(公告)号：US08131659B2

公开(公告)日：2012-03-06

申请号：US12238239

申请日：2008-09-25

申请人： Ning-Yi Xu ,  Xiong-Fei Cai ,  Rui Gao ,  Jing Yan ,  Feng-Hsiung Hsu

发明人： Ning-Yi Xu ,  Xiong-Fei Cai ,  Rui Gao ,  Jing Yan ,  Feng-Hsiung Hsu

IPC分类号： G06Q30/00

CPC分类号： G06N3/063

摘要： Accelerator systems and methods are disclosed that utilize FPGA technology to achieve better parallelism and processing speed. A Field Programmable Gate Array (FPGA) is configured to have a hardware logic performing computations associated with a neural network training algorithm, especially a Web relevance ranking algorithm such as LambaRank. The training data is first processed and organized by a host computing device, and then streamed to the FPGA for direct access by the FPGA to perform high-bandwidth computation with increased training speed. Thus, large data sets such as that related to Web relevance ranking can be processed. The FPGA may include a processing element performing computations of a hidden layer of the neural network training algorithm. Parallel computing may be realized using a single instruction multiple data streams (SIMD) architecture with multiple arithmetic logic units in the FPGA.

摘要翻译： 公开了利用FPGA技术实现更好的并行性和处理速度的加速器系统和方法。 现场可编程门阵列（FPGA）被配置为具有执行与神经网络训练算法相关联的计算的硬件逻辑，特别是诸如LambaRank的Web相关性排序算法。 训练数据首先由主机计算机处理和组织，然后流式传输到FPGA，以便FPGA直接访问，以提高训练速度进行高带宽计算。 因此，可以处理与Web相关性排名相关的大数据集。 FPGA可以包括执行神经网络训练算法的隐藏层的计算的处理元件。 可以使用FPGA中具有多个算术逻辑单元的单指令多数据流（SIMD）架构来实现并行计算。

公开(公告)号：US07946921B2

公开(公告)日：2011-05-24

申请号：US11134950

申请日：2005-05-23

申请人： Eyal Ofek ,  Feng-Hsiung Hsu

发明人： Eyal Ofek ,  Feng-Hsiung Hsu

IPC分类号： A63F13/00

CPC分类号： A63F13/213 ,  A63F13/06 ,  A63F13/42 ,  A63F13/92 ,  A63F2300/1093

摘要： In an exemplary embodiment, an on-board camera is used as a controller for a hand-held game device. Control information is provided to a game application via video captured by the camera. As the hand-held device is moved, features of the captured video are tracked. The relative motion between the tracked features and the hand-held device is translated into control information for the game application. Aspects of the game application are rendered on a display of the hand-held game device in accordance with the control information. For example, from a third person perspective, viewing a game character from overhead, rotation of the hand-held game device can be translated into rotation of the character. From a first person perspective, seeing the game through the eyes of the game character, rotation of the hand-held game device can be translated into observing various challenges coming from different directions.

摘要翻译： 在示例性实施例中，车载照相机用作手持游戏装置的控制器。 通过摄像机拍摄的视频将控制信息提供给游戏应用程序。 当手持设备移动时，捕获的视频的特征被跟踪。 跟踪功能和手持设备之间的相对运动被转换为游戏应用程序的控制信息。 游戏应用的方面根据控制信息呈现在手持游戏装置的显示器上。 例如，从第三人的角度来看，从头顶观看游戏角色，手持游戏装置的旋转可以转换为角色的旋转。 从第一人称角度来看，通过游戏角色观看游戏，可以将手持游戏装置的旋转转化为观察来自不同方向的各种挑战。

公开(公告)号：US07852315B2

公开(公告)日：2010-12-14

申请号：US11278998

申请日：2006-04-07

申请人： Eyal Ofek ,  Feng-Hsiung Hsu

发明人： Eyal Ofek ,  Feng-Hsiung Hsu

IPC分类号： G06F3/033 ,  G09G5/08

CPC分类号： G06F3/0346 ,  G06F3/0304

摘要： Systems and methods for a camera and acceleration based interface for presentations are described. In one aspect, a handheld device is utilized capture optical information. The optical information is then used to identify the location of a cursor that is presented on a display surface. While a user moves the handheld device, inertial information corresponding to the movement is accumulated on the handheld device. The optical and accumulated inertial information is used to modify position of the cursor with respect to the display surface.

摘要翻译： 描述用于相机的基于加速度的接口用于演示的系统和方法。 在一个方面，使用手持设备捕获光学信息。 然后，光学信息用于识别呈现在显示表面上的光标的位置。 当用户移动手持设备时，对应于移动的惯性信息被累积在手持设备上。 光学和累积的惯性信息用于修改光标相对于显示表面的位置。

公开(公告)号：US20100201808A1

公开(公告)日：2010-08-12

申请号：US12367665

申请日：2009-02-09

申请人： Feng-Hsiung Hsu

发明人： Feng-Hsiung Hsu

IPC分类号： H04N7/18

CPC分类号： A63F13/213 ,  A63F2300/1087 ,  A63F2300/8011 ,  G06F3/0325 ,  G06T7/73 ,  G06T2207/30204

摘要： Providing camera based motion detection is disclosed herein. A camera may track a reference element array of retroreflectors, which reflect light from a light source located proximate the camera. A known arrangement of the reference element array may be compared to a received arrangement on an image sensor of the camera to determine position information of at least one of the camera or the reference element array. The reference elements may include a style (pattern, shape) that may enable extraction of additional position information when the reference element array is captured by the camera. A user-manipulated device may be configured with the camera and light source, or alternatively, the reference elements, to enable communication with a computing device and display device.

摘要翻译： 本文公开了提供基于摄像机的运动检测。 相机可以跟踪反射器的参考元件阵列，其反射来自位于相机附近的光源的光。 可以将参考元件阵列的已知布置与相机的图像传感器上的接收布置进行比较，以确定相机或参考元件阵列中的至少一个的位置信息。 参考元件可以包括样式（图案，形状），当参考元件阵列被照相机捕获时，可以使得能够提取附加位置信息。 用户操纵的设备可以配置有相机和光源，或者替代地，参考元件，以实现与计算设备和显示设备的通信。

公开(公告)号：US20100076915A1

公开(公告)日：2010-03-25

申请号：US12238239

申请日：2008-09-25

申请人： Ning-Yi Xu ,  Xiong-Fei Cai ,  Rui Gao ,  Jing Yan ,  Feng-Hsiung Hsu

发明人： Ning-Yi Xu ,  Xiong-Fei Cai ,  Rui Gao ,  Jing Yan ,  Feng-Hsiung Hsu

IPC分类号： G06N3/08

CPC分类号： G06N3/063

摘要： Accelerator systems and methods are disclosed that utilize FPGA technology to achieve better parallelism and processing speed. A Field Programmable Gate Array (FPGA) is configured to have a hardware logic performing computations associated with a neural network training algorithm, especially a Web relevance ranking algorithm such as LambaRank. The training data is first processed and organized by a host computing device, and then streamed to the FPGA for direct access by the FPGA to perform high-bandwidth computation with increased training speed. Thus, large data sets such as that related to Web relevance ranking can be processed. The FPGA may include a processing element performing computations of a hidden layer of the neural network training algorithm. Parallel computing may be realized using a single instruction multiple data streams (SIMD) architecture with multiple arithmetic logic units in the FPGA.

摘要翻译： 公开了利用FPGA技术实现更好的并行性和处理速度的加速器系统和方法。 现场可编程门阵列（FPGA）被配置为具有执行与神经网络训练算法相关联的计算的硬件逻辑，特别是诸如LambaRank的Web相关性排序算法。 训练数据首先由主机计算机处理和组织，然后流式传输到FPGA，以便FPGA直接访问，以提高训练速度进行高带宽计算。 因此，可以处理与Web相关性排名相关的大数据集。 FPGA可以包括执行神经网络训练算法的隐藏层的计算的处理元件。 可以使用FPGA中具有多个算术逻辑单元的单指令多数据流（SIMD）架构来实现并行计算。

公开(公告)号：US07477236B2

公开(公告)日：2009-01-13

申请号：US11118720

申请日：2005-04-29

申请人： Eyal Ofek ,  Feng-Hsiung Hsu

发明人： Eyal Ofek ,  Feng-Hsiung Hsu

IPC分类号： G06F3/033 ,  G09G5/08

CPC分类号： G06F3/0346 ,  G06F3/0304 ,  G06F3/038

摘要： Systems and methods for remote control of on-screen interaction are described. In one aspect, a presentation is displayed onto a display surface. The presentation includes a selection control to interface with one or more user interface (UI) elements. The UI elements are used to control the presentation. A set of image frames are received from a remote handheld device. The image frames represent respective snapshots of the presentation. The image frames are evaluated to modify position of the selection control over the presentation.

摘要翻译： 描述用于远程控制屏幕交互的系统和方法。 在一个方面，将显示显示在显示表面上。 呈现包括与一个或多个用户界面（UI）元素接口的选择控件。 UI元素用于控制演示。 从远程手持设备接收一组图像帧。 图像帧表示呈现的各个快照。 评估图像帧以修改选择控件在演示文稿上的位置。

公开(公告)号：US20090009469A1

公开(公告)日：2009-01-08

申请号：US11774515

申请日：2007-07-06

申请人： Feng-Hsiung Hsu ,  Rui Gao

发明人： Feng-Hsiung Hsu ,  Rui Gao

IPC分类号： G06F3/033

CPC分类号： G06F3/0325

摘要： Motion-based control of an electronic device uses an array of at least three reference elements forming a triangle. An image sensor (e.g., a video camera), which may be located on a user-manipulated device, captures an image of the array. The array image has a pattern formed by a nonparallel projection of the reference triangle onto the image sensor. The pattern carries information of the relative position between the image sensor and the reference element array, and changes as the relative position changes. The pattern is identified and used for generating position information, which may express a multidimensional position of the user-manipulated device with respect to three axes describing a translational position, and three rotational axes describing pitch, roll and yaw motions. The control system and method are particularly suitable for videogames.

摘要翻译： 电子设备的基于运动的控制使用形成三角形的至少三个参考元素的阵列。 可以位于用户操纵的设备上的图像传感器（例如，摄像机）捕获阵列的图像。 阵列图像具有通过参考三角形到图像传感器的非平行投影形成的图案。 该图案携带图像传感器和参考元件阵列之间的相对位置的信息，并且随着相对位置的变化而改变。 该图案被识别并用于生成位置信息，其可以表示用户操纵的装置相对于描述平移位置的三个轴的多维位置，以及描述俯仰，滚动和偏航运动的三个旋转轴。 控制系统和方法特别适用于电子游戏。

公开(公告)号：US07425070B2

公开(公告)日：2008-09-16

申请号：US11128678

申请日：2005-05-13

申请人： Feng-Hsiung Hsu

发明人： Feng-Hsiung Hsu

IPC分类号： G03B21/14

CPC分类号： G03B35/20 ,  G03B21/625 ,  G03B35/24 ,  H04N13/302 ,  H04N13/351 ,  H04N13/363

摘要： Three dimensional (3D) image projection is described. In one aspect, each projector of multiple projectors generate a respective composite of an image. With respect to any particular projector, the projector's corresponding composite image encapsulates multiple vertical slices of the image. Each vertical slice in the composite image represents a particular portion of the image for projection by the projector onto a specific area of multiple areas on a screen. The size of the specific area is a function of a width of a viewpoint of multiple viewpoints, wherein the viewpoint is linearly positioned with the specific area on the screen and the projector.

摘要翻译： 描述三维（3D）图像投影。 在一个方面，多个投影仪的每个投影仪产生图像的相应复合物。 对于任何特定的投影机，投影仪的相应复合图像封装了图像的多个垂直切片。 合成图像中的每个垂直切片表示图像的特定部分，用于将投影仪投影到屏幕上多个区域的特定区域上。 特定区域的大小是多个视点的视点的宽度的函数，其中视点与屏幕上的特定区域和投影仪线性定位。

公开(公告)号：US07262700B2

公开(公告)日：2007-08-28

申请号：US11128510

申请日：2005-05-13

申请人： Feng-Hsiung Hsu

发明人： Feng-Hsiung Hsu

IPC分类号： G08B13/14

CPC分类号： H02J7/025 ,  H02J5/005 ,  H02J50/40 ,  H02J50/80

摘要： An inductive powering device provides power to a portable device via inductive coupling between primary coils in the surface of the powering device and a secondary coil in the portable device. The portable device includes a passive locator device, such as an RFID device, to allow the primary coils of the inductive powering surface to detect the presence and location of the secondary coil, and only primary coils adjacent the secondary coil are energized for power transfer. A cost-effective driving configuration that arranges the primary coils into a matrix with drive circuits switchably connected to the row and columns is used to energize the primary coils.

摘要翻译： 感应供电装置通过电源装置的表面中的初级线圈和便携式装置中的次级线圈之间的感应耦合向便携式装置提供电力。 便携式装置包括诸如RFID装置的无源定位装置，以允许感应供电表面的初级线圈检测次级线圈的存在和位置，并且只有邻近次级线圈的初级线圈被激励才能进行功率传递。 使用将初级线圈布置成矩阵的成本有效的驱动配置，驱动电路可切换地连接到行和列，以使初级线圈通电。