中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

2 records (took 0.016 seconds)

    REDUCED COMPUTATIONAL COMPLEXITY FOR FIXED POINT NEURAL NETWORK
    1.
    发明申请
    REDUCED COMPUTATIONAL COMPLEXITY FOR FIXED POINT NEURAL NETWORK 审中-公开
    固定点神经网络的降低计算复杂度

    公开(公告)号：US20160328645A1

    公开(公告)日：2016-11-10

    申请号：US14882351

    申请日：2015-10-13

    Applicant: QUALCOMM Incorporated

    Inventor： Dexu LIN ,  Matthew BADIN ,  David Edward HOWARD ,  Daniel Hendricus Franciscus DIJKMAN ,  Michael Colin TREMAINE ,  Anthony SARAH

    IPC: G06N3/08 ,  G06N99/00

    CPC classification number: G06N3/08 ,  G06N3/063 ,  G06N20/00

    Abstract: A method of reducing computational complexity for a fixed point neural network operating in a system having a limited bit width in a multiplier-accumulator (MAC) includes reducing a number of bit shift operations when computing activations in the fixed point neural network. The method also includes balancing an amount of quantization error and an overflow error when computing activations in the fixed point neural network.

    Abstract translation: 一种降低在乘法器 - 累加器（MAC）中具有有限位宽度的系统中操作的固定点神经网络的计算复杂度的方法包括在计算固定点神经网络中的激活时减少多个位移操作。 该方法还包括在计算固定点神经网络中的激活时平衡量化误差量和溢出误差。

    CONVOLUTION MATRIX MULTIPLY WITH CALLBACK FOR DEEP TILING FOR DEEP CONVOLUTIONAL NEURAL NETWORKS
    2.
    发明申请
    CONVOLUTION MATRIX MULTIPLY WITH CALLBACK FOR DEEP TILING FOR DEEP CONVOLUTIONAL NEURAL NETWORKS 审中-公开
    用于深呼吸神经网络的深呼吸调音矩阵

    公开(公告)号：US20160239706A1

    公开(公告)日：2016-08-18

    申请号：US14845243

    申请日：2015-09-03

    Applicant: QUALCOMM Incorporated

    Inventor： Daniel Hendricus Franciscus DIJKMAN ,  Matthew BADIN

    IPC: G06K9/00 ,  G06K9/66 ,  G06F17/30

    CPC classification number: G06K9/00503 ,  G06F16/51 ,  G06K9/4628 ,  G06K9/66 ,  G06N3/0454

    Abstract: A method of address translation of images and filters to virtual matrices to perform a convolution by matrix multiplication includes receiving an image and a filter. Each image and filter has a memory address. The method also includes mapping the memory addresses to virtual matrix addresses based on a calculated linearized image and a calculated linearized filter. The method further includes converting data in the virtual matrix to a predefined internal format. The method still further includes convolving the image by matrix multiplication of the data in the predefined internal format based on the virtual matrix addresses.

    Abstract translation: 将图像和滤波器地址转换为虚拟矩阵以通过矩阵乘法执行卷积的方法包括接收图像和滤波器。 每个图像和滤镜都有一个内存地址。 该方法还包括基于计算的线性化图像和计算的线性化滤波器将存储器地址映射到虚拟矩阵地址。 该方法还包括将虚拟矩阵中的数据转换成预定义的内部格式。 该方法还包括通过基于虚拟矩阵地址以预定内部格式的数据的矩阵乘法来卷积图像。

Patent Agency Ranking