公开(公告)号：WO2015020802A3

公开(公告)日：2015-05-14

申请号：PCT/US2014047858

申请日：2014-07-23

Applicant: QUALCOMM INC

Inventor： RANGAN VENKAT

IPC: G06N3/04 ,  G06N3/063 ,  G06N3/08

CPC classification number: G06N3/08 ,  G06N3/049 ,  G06N3/063 ,  G06N3/082

Abstract: Methods and apparatus are provided for determining synapses in an artificial nervous system based on connectivity patterns. One example method generally includes determining, for an artificial neuron, an event has occurred; based on the event, determining one or more synapses with other artificial neurons based on a connectivity pattern associated with the artificial neuron; and applying a spike from the artificial neuron to the other artificial neurons based on the determined synapses. In this manner, the connectivity patterns (or parameters for determining such patterns) for particular neuron types, rather than the connectivity itself, may be stored. Using the stored information, synapses may be computed on the fly, thereby reducing memory consumption and increasing memory bandwidth. This also saves time during artificial nervous system updates.

Abstract translation: 提供了用于基于连接模式确定人造神经系统中的突触的方法和装置。 一个示例性方法通常包括为人造神经元确定已经发生事件; 基于事件，基于与人造神经元相关联的连接模式，确定与其他人造神经元的一个或多个突触; 并根据确定的突触将人造神经元的尖峰应用于其他人造神经元。 以这种方式，可以存储用于特定神经元类型而不是连接本身的连接模式（或用于确定这种模式的参数）。 使用存储的信息，可以即时计算突触，从而减少内存消耗并增加内存带宽。 这也节省了人造神经系统更新中的时间。

公开(公告)号：WO2014025619A2

公开(公告)日：2014-02-13

申请号：PCT/US2013053290

申请日：2013-08-01

Applicant: QUALCOMM INC

Inventor： MAJUMDAR SOMDEB ,  RANGAN VENKAT ,  LEVIN JEFFREY A

IPC: G06N3/04

CPC classification number: G06N3/02 ,  G06N3/049 ,  G10L19/038 ,  G10L19/12

Abstract: Certain aspects of the present disclosure support a technique for optimized representation of variables in neural systems. Bit-allocation for neural signals and parameters in a neural network described in the present disclosure may comprise allocating quantization levels to the neural signals based on at least one measure of sensitivity of a pre-determined performance metric to quantization errors in the neural signals, and allocating bits to the parameters based on the at least one measure of sensitivity of the pre-determined performance metric to quantization errors in the parameters.

Abstract translation: 本公开的某些方面支持用于神经系统中变量的优化表示的技术。 在本公开中描述的神经网络中的神经信号和参数的位分配可以包括基于对神经信号中的量化误差的预定性能度量的灵敏度的至少一个度量来为神经信号分配量化级别，以及 基于对所述参数中的量化误差的所述预定性能度量的灵敏度的至少一个度量来将比特分配给所述参数。

公开(公告)号：WO2015034640A3

公开(公告)日：2015-05-14

申请号：PCT/US2014051044

申请日：2014-08-14

Applicant: QUALCOMM INC

Inventor： CANOY MICHAEL-DAVID NAKAYOSHI ,  BELL II WILLIAM RICHARD ,  KINTADA RAMAKRISHNA ,  RANGAN VENKAT

IPC: G06N3/04 ,  G06F11/30 ,  G06F11/36 ,  G06N3/06 ,  G06N3/08 ,  G06N3/10

CPC classification number: G06N3/10 ,  G06F11/302 ,  G06F11/3636 ,  G06N3/049 ,  G06N3/08

Abstract: Methods and apparatus are provided for using a breakpoint determination unit to examine an artificial nervous system. One example method generally includes operating at least a portion of the artificial nervous system; using the breakpoint determination unit to detect that a condition exists based at least in part on monitoring one or more components in the artificial nervous system; and at least one of suspending, examining, modifying, or flagging the operation of the at least the portion of the artificial nervous system, based at least in part on the detection.

Abstract translation: 提供了使用断点确定单元检查人造神经系统的方法和设备。 一个示例性方法通常包括操作人造神经系统的至少一部分; 使用所述断点确定单元至少部分地基于监测所述人造神经系统中的一个或多个组件来检测所述条件存在; 以及至少部分基于所述检测来暂停，检查，修改或标记所述人造神经系统的所述至少一部分的操作中的至少一者。

公开(公告)号：WO2014189970A3

公开(公告)日：2015-04-09

申请号：PCT/US2014038841

申请日：2014-05-20

Applicant: QUALCOMM INC

Inventor： LEVIN JEFFREY ALEXANDER ,  RANGAN VENKAT ,  MALONE ERIK CHRISTOPHER

IPC: G06N3/04 ,  G06N3/06

CPC classification number: G06N3/063 ,  G06N3/049 ,  G06N3/08

Abstract: Certain aspects of the present disclosure support operating simultaneously multiple super neuron processing units in an artificial nervous system, wherein a plurality of artificial neurons is assigned to each super neuron processing unit. The super neuron processing units can be interfaced with a memory for storing and loading synaptic weights and plasticity parameters of the artificial nervous system, wherein organization of the memory allows contiguous memory access.

Abstract translation: 本公开的某些方面支持在人造神经系统中同时操作多个超级神经元处理单元，其中将多个人造神经元分配给每个超级神经元处理单元。 超神经元处理单元可以与用于存储和加载人造神经系统的突触权重和可塑性参数的存储器连接，其中存储器的组织允许连续的存储器访问。

公开(公告)号：WO2015053889A3

公开(公告)日：2015-06-11

申请号：PCT/US2014054510

申请日：2014-09-08

Applicant: QUALCOMM INC

Inventor： RANGAN VENKAT ,  WEGRZYN JAN KRZYS ,  LEVIN JEFFREY ALEXANDER ,  DANIELS JOHN PAUL

IPC: G06N3/04 ,  G06N3/063

CPC classification number: G06F12/02 ,  G06N3/049 ,  G06N3/063

Abstract: Aspects of the present disclosure provide methods and apparatus for allocating memory in an artificial nervous system simulator implemented in hardware. According to certain aspects, memory resource requirements for one or more components of an artificial nervous system being simulated may be determined and portions of a shared memory pool (which may include on-chip and/or off-chip RAM) may be allocated to the components based on the determination.

Abstract translation: 本公开的各方面提供了用于在以硬件实现的人造神经系统模拟器中分配存储器的方法和设备。 根据某些方面，可以确定正被模拟的人造神经系统的一个或多个组件的存储器资源需求，并且可以将共享存储器池（其可以包括片上和/或片外RAM）的部分分配给 基于确定的组件。

公开(公告)号：WO2014197175A2

公开(公告)日：2014-12-11

申请号：PCT/US2014038008

申请日：2014-05-14

Applicant: QUALCOMM INC

Inventor： RANGAN VENKAT

IPC: G06N3/04

CPC classification number: G06N3/063 ,  G06N3/049 ,  G06N3/10

Abstract: Certain aspects of the present disclosure support a technique for efficient implementation of neural population diversity in neural systems. A set of parameters for each class of artificial neurons of a plurality of classes can be stored in a storage medium. A generator can be configured to obtain noise parameters for each class of artificial neurons in the neural system. After that, the noise parameters can be combined with the set of parameters for each class of artificial neurons to obtain a dithered set of parameters for each class of artificial neurons. The dithered set of parameters can be stored for each class of artificial neurons to be used for a neuron model for the artificial neurons that emulates behavior of the artificial neurons in the neural system.

Abstract translation: 本公开的某些方面支持用于神经系统中有效实施神经群体多样性的技术。 可以将多组类别的每类人造神经元的一组参数存储在存储介质中。 发电机可以被配置为获得神经系统中每一类人造神经元的噪声参数。 之后，可以将噪声参数与每类人造神经元的参数组合，以获得每类人造神经元的抖动参数集合。 可以为每类人造神经元存储抖动的参数集合，以用于仿真神经系统中人造神经元行为的人造神经元的神经元模型。

公开(公告)号：WO2015020815A3

公开(公告)日：2015-06-11

申请号：PCT/US2014048187

申请日：2014-07-25

Applicant: QUALCOMM INC

Inventor： MALONE ERIK CHRISTOPHER ,  RANGAN VENKAT ,  LEVIN JEFFREY ALEXANDER

IPC: G06N3/04

CPC classification number: G06N3/02 ,  G06N3/049

Abstract: Methods and apparatus are provided for implementing delays in an artificial nervous system. Synaptic and/or axonal delays between a post-synaptic artificial neuron and one or more pre-synaptic artificial neurons may be accounted for at the post-synaptic artificial neuron. One example method for managing delay between neurons in an artificial nervous system generally includes receiving, at a post-synaptic artificial neuron, input current values from one or more pre-synaptic artificial neurons; accounting for delays between the one or more pre-synaptic artificial neurons and the post-synaptic artificial neuron at the post-synaptic artificial neuron; and determining a state of the post-synaptic artificial neuron based at least in part on at least a portion of the input current values, according to the accounting.

Abstract translation: 提供了用于实施人造神经系统中的延迟的方法和设备。 在突触后人工神经元和一个或多个突触前人工神经元之间的突触和/或轴突延迟可以在突触后人工神经元处考虑。 用于管理人造神经系统中的神经元之间的延迟的一个示例方法一般包括在突触后人工神经元处接收来自一个或多个突触前人造神经元的输入电流值; 说明突触后人造神经元的一个或多个突触前人工神经元与突触后人工神经元之间的延迟; 以及根据所述计算至少部分地基于所述输入电流值的至少一部分来确定所述突触后人造神经元的状态。

公开(公告)号：WO2014025621A2

公开(公告)日：2014-02-13

申请号：PCT/US2013053293

申请日：2013-08-01

Applicant: QUALCOMM INC

Inventor： RANGAN VENKAT ,  LEVIN JEFFREY A ,  BERAHA RODOLFO G

IPC: H04L12/26

CPC classification number: H04L43/106 ,  G06F15/7825 ,  H04L41/0659 ,  H04L41/0668 ,  H04L43/04 ,  H04L43/0852 ,  H04L43/10 ,  H04L43/50

Abstract: Certain aspects of the present disclosure support techniques for collecting system information in a network on a chip (NoC). A dedicated packet may be transmitted from a source node to a destination node. As it traverses through the NoC, the dedicated packet may collect information from various nodes, which may be made available by the destination node. The collected information may be used in an effort to detect failures and collect statistics regarding the NoC.

Abstract translation: 本公开的某些方面支持用于在芯片上的网络（NoC）中​​收集系统信息的技术。 专用分组可以从源节点发送到目的地节点。 当它通过NoC时，专用分组可以从各个节点收集信息，这些信息可能由目的地节点可用。 收集的信息可用于检测故障并收集有关NoC的统计信息。