公开(公告)号：US09984337B2

公开(公告)日：2018-05-29

申请号：US14872521

申请日：2015-10-01

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Erik Kruus ,  Hao Li

IPC: G06F17/30 ,  G06N99/00 ,  G06F9/46

CPC classification number: G06N99/005 ,  G06F9/46 ,  G06F17/30174 ,  G06F17/30209 ,  G06F17/30578

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.

公开(公告)号：US20170337471A1

公开(公告)日：2017-11-23

申请号：US15590620

申请日：2017-05-09

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Igor Durdanovic ,  Hans Peter Graf ,  Hao Li

IPC: G06N3/08 ,  G06N3/04 ,  H03H21/00 ,  G06N3/063 ,  G06N3/06

CPC classification number: G06N3/082 ,  G06F17/153 ,  G06F17/17 ,  G06K9/00771 ,  G06K9/4628 ,  G06K9/6228 ,  G06K9/627 ,  G06K9/6296 ,  G06K9/66 ,  G06K2009/00738 ,  G06N3/0427 ,  G06N3/0454 ,  G06N3/0481 ,  G06N5/045 ,  G08B13/00 ,  G08B29/186 ,  H03H2222/04

Abstract: Methods and systems for pruning a convolutional neural network (CNN) include calculating a sum of weights for each filter in a layer of the CNN. The filters in the layer are sorted by respective sums of weights. A set of m filters with the smallest sums of weights is filtered to decrease a computational cost of operating the CNN. The pruned CNN is retrained to repair accuracy loss that results from pruning the filters.

公开(公告)号：US20160103901A1

公开(公告)日：2016-04-14

申请号：US14872521

申请日：2015-10-01

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Erik Kruus ,  Hao Li

IPC: G06F17/30 ,  G06N99/00 ,  H04L12/751

CPC classification number: G06N99/005 ,  G06F9/46 ,  G06F17/30174 ,  G06F17/30209 ,  G06F17/30578

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。

公开(公告)号：US20170337467A1

公开(公告)日：2017-11-23

申请号：US15590666

申请日：2017-05-09

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Igor Durdanovic ,  Hans Peter Graf ,  Hao Li

IPC: G06N3/04 ,  G06K9/62 ,  G06F17/17 ,  G06K9/66 ,  G06F17/15

CPC classification number: G06N3/082 ,  G06F17/153 ,  G06F17/17 ,  G06K9/00771 ,  G06K9/4628 ,  G06K9/6228 ,  G06K9/627 ,  G06K9/6296 ,  G06K9/66 ,  G06K2009/00738 ,  G06N3/0427 ,  G06N3/0454 ,  G06N3/0481 ,  G06N5/045 ,  G08B13/00 ,  G08B29/186 ,  H03H2222/04

Abstract: Security systems and methods for detecting intrusion events include one or more sensors configured to monitor an environment. A pruned convolutional neural network (CNN) is configured process information from the one or more sensors to classify events in the monitored environment. CNN filters having the smallest summed weights have been pruned from the pruned CNN. An alert module is configured to detect an intrusion event in the monitored environment based on event classifications. A control module is configured to perform a security action based on the detection of an intrusion event.

公开(公告)号：US10832136B2

公开(公告)日：2020-11-10

申请号：US15590620

申请日：2017-05-09

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Igor Durdanovic ,  Hans Peter Graf ,  Hao Li

IPC: G06N3/08 ,  G06N3/04 ,  G06K9/62 ,  G06K9/00 ,  G06K9/46 ,  G08B13/00 ,  G08B29/18 ,  G06F17/15 ,  G06F17/17 ,  G06K9/66 ,  G06N5/04

Abstract: Methods and systems for pruning a convolutional neural network (CNN) include calculating a sum of weights for each filter in a layer of the CNN. The filters in the layer are sorted by respective sums of weights. A set of m filters with the smallest sums of weights is filtered to decrease a computational cost of operating the CNN. The pruned CNN is retrained to repair accuracy loss that results from pruning the filters.

公开(公告)号：US10885437B2

公开(公告)日：2021-01-05

申请号：US15590666

申请日：2017-05-09

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Igor Durdanovic ,  Hans Peter Graf ,  Hao Li

IPC: G06N3/04 ,  G06N3/08 ,  G06K9/62 ,  G06K9/00 ,  G06K9/46 ,  G08B13/00 ,  G08B29/18 ,  G06F17/15 ,  G06F17/17 ,  G06K9/66 ,  G06N5/04

Abstract: Security systems and methods for detecting intrusion events include one or more sensors configured to monitor an environment. A pruned convolutional neural network (CNN) is configured process information from the one or more sensors to classify events in the monitored environment. CNN filters having the smallest summed weights have been pruned from the pruned CNN. An alert module is configured to detect an intrusion event in the monitored environment based on event classifications. A control module is configured to perform a security action based on the detection of an intrusion event.

公开(公告)号：US20160125316A1

公开(公告)日：2016-05-05

申请号：US14875773

申请日：2015-10-06

Applicant: NEC Laboratories America, Inc.

Inventor： Asim Kadav ,  Cristian Ungureanu ,  Erik Kruus ,  Hao Li

IPC: G06N99/00

CPC classification number: G06N20/00 ,  G06F9/46 ,  G06F16/178 ,  G06F16/1837 ,  G06F16/273

Abstract: Systems and methods are disclosed for parallel machine learning with a cluster of N parallel machine network nodes by determining k network nodes as a subset of the N network nodes to update learning parameters, wherein k is selected to disseminate the updates across all nodes directly or indirectly and to optimize predetermined goals including freshness, balanced communication and computation ratio in the cluster; sending learning unit updates to fewer nodes to reduce communication costs with learning convergence; and sending reduced learning updates and ensuring that the nodes send/receive learning updates in a uniform fashion.

Abstract translation: 通过将k个网络节点确定为N个网络节点的子集来更新学习参数，公开了用于并行机器网络节点的并行机器学习的系统和方法，其中k被选择以直接或间接地在所有节点上传播更新 并优化包括新鲜度，平衡的通信和群集中的计算比例的预定目标; 将学习单元更新发送到较少的节点，以通过学习融合降低通信成本; 并发送减少的学习更新，并确保节点以统一的方式发送/接收学习更新。