Parallelized Machine Learning With Distributed Lockless Training
    161.
    发明申请
    Parallelized Machine Learning With Distributed Lockless Training 有权
    并行机器学习与分布式无锁训练

    公开(公告)号:US20160103901A1

    公开(公告)日:2016-04-14

    申请号:US14872521

    申请日:2015-10-01

    Abstract: Systems and methods are disclosed for providing distributed learning over a plurality of parallel machine network nodes by allocating a per-sender receive queue at every machine network node and performing distributed in-memory training; and training each unit replica and maintaining multiple copies of the unit replica being trained, wherein all unit replicas train, receive unit updates and merge in parallel in a peer-to-peer fashion, wherein each receiving machine network node merges updates at later point in time without interruption and wherein the propagating and synchronizing unit replica updates are lockless and asynchronous.

    Abstract translation: 公开了用于通过在每个机器网络节点处分配每发送器接收队列并执行分布式存储器内训练来在多个并行机器网络节点上提供分布式学习的系统和方法; 并训练每个单元复制品并维护被训练的单元副本的多个副本,其中所有单元副本训练,接收单元更新并且以对等方式并行并入,其中每个接收机网络节点在稍后的点处合并更新 时间不间断,其中传播和同步单元副本更新是无锁定和异步的。

    Integrated approach to model time series dynamics in complex physical systems
    162.
    发明授权
    Integrated approach to model time series dynamics in complex physical systems 有权
    在复杂物理系统中建立时间序列动力学的综合方法

    公开(公告)号:US09245235B2

    公开(公告)日:2016-01-26

    申请号:US14050945

    申请日:2013-10-10

    Abstract: A system and method for analysis of complex systems which includes determining model parameters based on time series data, further including profiling a plurality of types of data properties to discover complex data properties and dependencies; classifying the data dependencies into predetermined categories for analysis; and generating a plurality of models based on the discovered properties and dependencies. The system and method may analyze, using a processor, the generated models based on a fitness score determined for each model to generate a status report for each model; integrate the status reports for each model to determine an anomaly score for the generated models; and generate an alarm when the anomaly score exceeds a predefined threshold.

    Abstract translation: 一种用于分析复杂系统的系统和方法,包括基于时间序列数据确定模型参数,还包括分析多种类型的数据属性以发现复杂数据属性和依赖性; 将数据依赖关系分类为预定类别进行分析; 以及基于所发现的属性和依赖关系生成多个模型。 系统和方法可以基于为每个模型确定的适应度分数来使用处理器分析生成的模型,以生成每个模型的状态报告; 整合每个模型的状态报告,以确定生成的模型的异常得分; 并且当异常得分超过预定阈值时产生报警。

    Combining I-Q and/or PolMux optical receiver to enable single detector
    163.
    发明授权
    Combining I-Q and/or PolMux optical receiver to enable single detector 有权
    结合I-Q和/或PolMux光接收器,实现单检测

    公开(公告)号:US09203521B2

    公开(公告)日:2015-12-01

    申请号:US14457460

    申请日:2014-08-12

    CPC classification number: H04B10/613 H04B10/612 H04B10/614

    Abstract: A method for reducing optical components at a receiver which include converting an input signal at a receiver to include an interleaving of alternate signal diversity components, the signal diversity components including phase diversity when the converting includes 0 and 90 degree interleaving and the signal diversity components include polarization diversity interleaving when the converting includes interleaved orthogonal polarizations, and combining the signal diversity components for enabling a single photo detection at the receiver to detect the alternative signal diversity components for subsequent analog-to-digital conversion.

    Abstract translation: 一种用于减少接收机处的光学部件的方法,包括在接收机处转换输入信号以包括交替的信号分集分量的交织,所述信号分集分量包括当转换包括0和90度交织时的相位分集,并且信号分集分量包括 当转换包括交错的正交偏振时,偏振分集交错,以及组合信号分集分量,以使接收机处的单次光检测能够检测用于随后的模数转换的备选信号分集分量。

    Nonbinary LDPC coded modulation without bandwidth expansion for high-speed optical transmission
    167.
    发明授权
    Nonbinary LDPC coded modulation without bandwidth expansion for high-speed optical transmission 有权
    用于高速光传输的无带宽扩展的非二进制LDPC编码调制

    公开(公告)号:US09166728B2

    公开(公告)日:2015-10-20

    申请号:US13646392

    申请日:2012-10-05

    CPC classification number: H04J14/04 H04B10/5055 H04J14/06

    Abstract: Systems and methods for data transport, comprising encoding one or more streams of input data using nonbinary low density parity check (NB-LDPC) encoders, corresponding to orthogonal polarization states. Receiving one or more streams of input data using a buffer coupled to the encoders, the data written to the buffer bR bits at a time, where R is the code rate. Generating one or more signals using a 2b-ary mapper implemented as a look-up table (LUT) to store coordinates of a corresponding signal constellation, the 2b-ary mapper configured to assign bits of one or more signals to a signal constellation and to associate the bits of the signals with signal constellation points, wherein the constellation is expanded to avoid bandwidth expansion due to coding, generating substantial net coding gains within a same bandwidth. Modulating nonbinary LDPC-coded data streams using in-phase/quadrature (I/Q) modulators and multiplexing the data streams using polarization beam combiner.

    Abstract translation: 用于数据传输的系统和方法,包括使用对应于正交极化状态的非二进制低密度奇偶校验(NB-LDPC)编码器来编码一个或多个输入数据流。 使用耦合到编码器的缓冲器接收一个或多个输入数据流,每次写入缓冲器bR位的数据,其中R是码率。 使用实现为查找表(LUT)的2b-ary映射器来存储一个或多个信号来存储对应的信号星座的坐标,2b-ary映射器被配置为将一个或多个信号的位分配给信号星座,并且 将信号的位与信号星座点相关联,其中扩展星座以避免由于编码而导致的带宽扩展,从而在相同带宽内产生实质的网络编码增益。 使用同相/正交(I / Q)调制器调制非二进制LDPC编码数据流,并使用偏振束组合器复用数据流。

    Symmetric coherent OFDMA-PON configuration with low-speed ONU-side
    168.
    发明授权
    Symmetric coherent OFDMA-PON configuration with low-speed ONU-side 有权
    具有低速ONU侧的对称相干OFDMA-PON配置

    公开(公告)号:US09160480B2

    公开(公告)日:2015-10-13

    申请号:US13470425

    申请日:2012-05-14

    CPC classification number: H04J14/0282 H04J14/002

    Abstract: A method by an optical network unit ONU includes, for downstream transmission, using a first tunable laser for coherent detection on a sub-band basis to increase receiver sensitivity and reduce analog-to-digital conversion ADC and digital signal processor DSP requirements within the ONU, and for upstream transmission, using a second tunable laser and using an optical signal beating between the first and second tunable lasers to generate a tunable radio frequency RF signal source for upstream multi-band OFDMA signal generation thereby avoiding need for an otherwise more costly RF clock source within the ONU, enabling low-speed digital-to-analog conversion DAC operation and rendering the ONU colorless in both optical and radio frequency RF domains.

    Abstract translation: 光网络单元ONU的方法包括下行传输,使用第一可调谐激光器进行基于子带的相干检测,以增加接收机灵敏度并减少ONU内的模数转换ADC和数字信号处理器DSP的要求 并且对于上游传输,使用第二可调激光器并且使用在第一和第二可调谐激光器之间拍打的光信号来产生用于上行多频带OFDMA信号生成的可调谐射频信号源,从而避免需要另外更昂贵的RF ONU内的时钟源,实现低速数模转换DAC操作,并使ONU在光和射频领域都无色。

    Method for simultaneous scheduling of processes and offloading computation on many-core coprocessors
    169.
    发明授权
    Method for simultaneous scheduling of processes and offloading computation on many-core coprocessors 有权
    多核协同处理器同步进程调度和卸载计算方法

    公开(公告)号:US09152467B2

    公开(公告)日:2015-10-06

    申请号:US13858039

    申请日:2013-04-06

    CPC classification number: G06F9/5044 G06F9/5033 G06F2209/509

    Abstract: A method is disclosed to manage a multi-processor system with one or more manycore devices, by managing real-time bag-of-tasks applications for a cluster, wherein each task runs on a single server node, and uses the offload programming model, and wherein each task has a deadline and three specific resource requirements: total processing time, a certain number of manycore devices and peak memory on each device; when a new task arrives, querying each node scheduler to determine which node can best accept the task and each node scheduler responds with an estimated completion time and a confidence level, wherein the node schedulers use an urgency-based heuristic to schedule each task and its offloads; responding to an accept/reject query phase, wherein the cluster scheduler send the task requirements to each node and queries if the node can accept the task with an estimated completion time and confidence level; and scheduling tasks and offloads using a aging and urgency-based heuristic, wherein the aging guarantees fairness, and the urgency prioritizes tasks and offloads so that maximal deadlines are met.

    Abstract translation: 公开了一种通过管理用于集群的实时任务应用程序来管理具有一个或多个管理器设备的多处理器系统的方法,其中每个任务在单个服务器节点上运行,并且使用卸载编程模型, 并且其中每个任务具有最后期限和三个特定的资源要求:总处理时间,一定数量的管理设备和每个设备上的峰值存储器; 当新任务到达时,查询每个节点调度器以确定哪个节点可以最好地接受任务,并且每个节点调度器以估计的完成时间和置信水平进行响应,其中节点调度器使用基于紧急度的启发式来安排每个任务及其 卸货 响应于接受/拒绝查询阶段,其中所述群集调度器向每个节点发送所述任务需求,并且查询所述节点是否可以接收具有估计完成时间和置信水平的所述任务; 并使用基于老化和紧急性的启发式调度任务和卸载,其中老化保证公平性,并且紧急性优先考虑任务和卸载,以便满足最大期限。

Patent Agency Ranking