公开(公告)号：WO2016197054A1

公开(公告)日：2016-12-08

申请号：PCT/US2016/035901

申请日：2016-06-03

申请人： GOOGLE INC.

发明人： DESJARDINS, Guillaume ,  SIMONYAN, Karen ,  KAVUKCUOGLU, Koray ,  PASCANU, Razvan

IPC分类号： G06N3/08 ,  G06N3/04

CPC分类号： G06N3/08 ,  G06N3/04 ,  G06N3/084 ,  G06N99/005

摘要： Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing inputs using a neural network system that includes a whitened neural network layer. One of the methods includes receiving an input activation generated by a layer before the whitened neural network layer in the sequence; processing the received activation in accordance with a set of whitening parameters to generate a whitened activation; processing the whitened activation in accordance with a set of layer parameters to generate an output activation; and providing the output activation as input to a neural network layer after the whitened neural network layer in the sequence.

摘要翻译： 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于使用包含白化神经网络层的神经网络系统处理输入。 其中一种方法包括在序列中接收由白化神经网络层之前的层产生的输入激活; 根据一组增白参数来处理接收到的激活以产生白化激活; 根据一组层参数处理白化激活以产生输出激活; 并且在序列中的白化神经网络层之后，将输出激活提供给神经网络层的输入。

公开(公告)号：WO2016160539A1

公开(公告)日：2016-10-06

申请号：PCT/US2016/024134

申请日：2016-03-25

申请人： EQUIFAX, INC.

发明人： TURNER, Matthew ,  MCBURNETT, Michael

IPC分类号： G06N3/04 ,  G06N3/08

CPC分类号： G06N3/084 ,  G06F15/18 ,  G06F17/18 ,  G06N3/04 ,  G06N3/082 ,  G06N7/005 ,  G06Q40/00 ,  G06Q40/025 ,  G06Q40/08

摘要： Certain embodiments involve generating or optimizing a neural network for risk assessment. The neural network can be generated using a relationship between various predictor variables and an outcome (e.g., a condition's presence or absence). The neural network can be used to determine a relationship between each of the predictor variables and a risk indicator. The neural network can be optimized by iteratively adjusting the neural network such that a monotonic relationship exists between each of the predictor variables and the risk indicator. The optimized neural network can be used both for accurately determining risk indicators using predictor variables and determining adverse action codes for the predictor variables, which indicate an effect or an amount of impact that a given predictor variable has on the risk indicator. The neural network can be used to generate adverse action codes upon which consumer behavior can be modified to improve the risk indicator score.

摘要翻译： 某些实施例涉及生成或优化用于风险评估的神经网络。 可以使用各种预测变量和结果之间的关系（例如，条件的存在或不存在）来生成神经网络。 神经网络可用于确定每个预测变量与风险指标之间的关系。 可以通过迭代调整神经网络来优化神经网络，使得每个预测变量和风险指标之间存在单调关系。 优化的神经网络可用于使用预测变量准确确定风险指标，并确定预测变量的不利行为代码，这些变量表示给定预测变量对风险指标的影响或影响量。 神经网络可以用于产生不利的行为代码，消费者行为可以被修改以改善风险指标得分。

公开(公告)号：WO2016053748A1

公开(公告)日：2016-04-07

申请号：PCT/US2015/051936

申请日：2015-09-24

申请人： SIKORSKY AIRCRAFT CORPORATION

发明人： GIERING, Michael J. ,  SHASHANKA, Madhusudana

IPC分类号： G06N3/08 ,  G06N3/04 ,  G01H17/00

CPC分类号： G06N3/084 ,  G01H1/00 ,  G01H1/003 ,  G06N3/0445 ,  G06N3/0472 ,  G06N3/0481 ,  G06N3/063 ,  G06N3/10

摘要： A system and method for providing health indication of a mechanical system, includes receiving signals indicative of vibration data of the mechanical system; pre-training features in the signals with a model; determining information related to vibration signatures in the signals; associating the vibration signatures with historical vibration data of the mechanical system; and building a multi-layer Deep Neural Network (DNN) from the vibration signatures and the historical vibration data.

摘要翻译： 一种用于提供机械系统的健康指示的系统和方法，包括接收指示所述机械系统的振动数据的信号; 模拟信号中的训练前特征; 确定信号中与振动特征相关的信息; 将振动特征与机械系统的历史振动数据相关联; 并从振动特征和历史振动数据建立多层深层神经网络（DNN）。

公开(公告)号：WO2015134900A1

公开(公告)日：2015-09-11

申请号：PCT/US2015/019236

申请日：2015-03-06

申请人： PROGRESS, INC.

发明人： PESCIANSCHI, Dmitri

IPC分类号： G06E1/00

CPC分类号： G06N3/08 ,  G06N3/04 ,  G06N3/084

摘要： A neural network includes a plurality of inputs for receiving input signals, and synapses connected to the inputs and having corrective weights. The network additionally includes distributors. Each distributor is connected to one of the inputs for receiving the respective input signal and selects one or more corrective weights in correlation with the input value. The network also includes neurons. Each neuron has an output connected with at least one of the inputs via one synapse and generates a neuron sum by summing corrective weights selected from each synapse connected to the respective neuron. Furthermore, the network includes a weight correction calculator that receives a desired output signal, determines a deviation of the neuron sum from the desired output signal value, and modifies respective corrective weights using the determined deviation. Adding up the modified corrective weights to determine the neuron sum minimizes the subject deviation for training the neural network.

摘要翻译： 神经网络包括用于接收输入信号的多个输入和连接到输入并具有校正权重的突触。 该网络还包括分销商。 每个分配器连接到一个输入端以接收相应的输入信号，并且与输入值相关地选择一个或多个校正权重。 该网络还包括神经元。 每个神经元具有通过一个突触与至少一个输入连接的输出，并且通过对从连接到各个神经元的每个突触中选择的校正权重相加来产生神经元和。 此外，网络包括权重校正计算器，其接收期望的输出信号，确定神经元和与期望输出信号值的偏差，并使用所确定的偏差修改相应的校正权重。 添加修改的校正权重以确定神经元总和使训练神经网络的主体偏差最小化。

公开(公告)号：WO2015103520A1

公开(公告)日：2015-07-09

申请号：PCT/US2015/010114

申请日：2015-01-05

申请人： CISCO TECHNOLGY, INC.

发明人： VASSEUR, Jean-Philippe ,  CRUZ MOTA, Javier ,  DI PIETRO, Andrea

IPC分类号： H04L12/24 ,  H04L29/06

CPC分类号： G06N99/005 ,  G06N3/0454 ,  G06N3/08 ,  G06N3/084 ,  H04L41/142 ,  H04L41/16 ,  H04L43/10 ,  H04L63/14 ,  H04L63/1408 ,  H04L63/1416

摘要： Training method to train machine learning model (e.g., neural network) used to identify attacks in a network (e.g., denial of service). A first data set is received by a network device that is indicative of the statuses of a plurality of network devices when a type of network attack is not present. A second data set is also received that is indicative of the statuses of the plurality of network devices when the type of network attack is present. At least one of the plurality simulates the type of network attack by operating as an attacking node. A machine learning model is trained using the first and second data set to identify the type of network attack. A real network attack is then identified using the trained machine learning model. Application in Low Power and Lossy Networks.

摘要翻译： 用于训练用于识别网络中的攻击的机器学习模型（例如，神经网络）的训练方法（例如，拒绝服务）。 当网络攻击的类型不存在时，由网络设备接收第一数据集，其指示多个网络设备的状态。 还接收当存在网络攻击的类型时指示多个网络设备的状态的第二数据集。 多个中的至少一个通过作为攻击节点操作来模拟网络攻击的类型。 使用第一和第二数据集来训练机器学习模型以识别网络攻击的类型。 然后使用训练有素的机器学习模型识别真实的网络攻击。 在低功耗和有损网络中的应用。

公开(公告)号：WO2015021016A1

公开(公告)日：2015-02-12

申请号：PCT/US2014/049724

申请日：2014-08-05

申请人： BOARD OF TRUSTEES OF THE UNIVERSITY OF ALABAMA ,  CITY UNIVERSITY OF LONDON ,  MISSOURI UNIVERSITY OF SCIENCE AND TECHNOLOGY

发明人： LI, Shuhui ,  FAIRBANK, Michael ,  FU, Xingang ,  WUNSCH, Donald ,  ALONSO, Eduardo

IPC分类号： H02P21/00

CPC分类号： G06N3/08 ,  G05B13/027 ,  G06N3/084

摘要： An example method for controlling an AC electrical machine can include providing a PWM converter operably connected between an electrical power source and the AC electrical machine and providing a neural network vector control system operably connected to the PWM converter. The control system can include a current-loop neural network configured to receive a plurality of inputs. The current-loop neural network can be configured to optimize the compensating dq-control voltage. The inputs can be d- and q-axis currents, d- and q-axis error signals, predicted d- and q-axis current signals, and a feedback compensating dq-control voltage. The d- and q-axis error signals can be a difference between the d- and q-axis currents and reference d- and q-axis currents, respectively. The method can further include outputting a compensating dq-control voltage from the current-loop neural network and controlling the PWM converter using the compensating dq-control voltage.

摘要翻译： 用于控制AC电机的示例性方法可以包括提供可操作地连接在电源和AC电机之间的PWM转换器，并提供可操作地连接到PWM转换器的神经网络矢量控制系统。 控制系统可以包括被配置为接收多个输入的电流环神经网络。 电流环神经网络可以配置为优化补偿dq控制电压。 输入可以是d轴和q轴电流，d轴和q轴误差信号，预测的d轴和q轴电流信号，以及反馈补偿dq控制电压。 d轴和q轴误差信号可分别为d轴和q轴电流与参考d轴和q轴电流之间的差值。 该方法还可以包括从电流环路神经网络输出补偿dq控制电压，并使用补偿dq控制电压来控制PWM转换器。

公开(公告)号：WO2014109847A1

公开(公告)日：2014-07-17

申请号：PCT/US2013/073374

申请日：2013-12-05

申请人： SENSORY, INCORPORATED

发明人： HOSOM, John-Paul ,  VERMEULEN, Pieter J. ,  SHAW, Johnathan

IPC分类号： G10L15/16 ,  G10L15/02 ,  G10L15/28

CPC分类号： G10L17/14 ,  G06N3/082 ,  G06N3/084 ,  G10L17/04 ,  G10L17/18

摘要： In one embodiment, a computer system stores speech data for a plurality of speakers, where the speech data includes a plurality of feature vectors and, for each feature vector, an associated sub-phonetic class. The computer system then builds, based on the speech data, an artificial neural network (ANN) for modeling speech of a target speaker in the plurality of speakers, where the ANN is configured to discriminate between instances of sub-phonetic classes uttered by the target speaker and instances of sub-phonetic classes uttered by other speakers in the plurality of speakers.

摘要翻译： 在一个实施例中，计算机系统存储用于多个扬声器的语音数据，其中语音数据包括多个特征向量，并且对于每个特征向量，存在相关联的子语音类。 计算机系统然后基于语音数据构建用于对多个扬声器中的目标扬声器的语音进行建模的人造神经网络（ANN），其中ANN被配置为区分由目标发出的子语音类别的实例 扬声器和多个扬声器中的其他扬声器发出的子语音类的实例。

公开(公告)号：WO2012036934A1

公开(公告)日：2012-03-22

申请号：PCT/US2011/050472

申请日：2011-09-06

申请人： MICROSOFT CORPORATION

发明人： DENG, Li ,  YU, Dong ,  DAHL, George Edward

IPC分类号： G10L15/14 ,  G10L15/16

CPC分类号： G10L15/14 ,  G06N3/0454 ,  G06N3/084

摘要： A method is disclosed herein that includes an act of causing a processor to receive a sample, wherein the sample is one of spoken utterance, an online handwriting sample, or a moving image sample. The method also comprises the act of causing the processor to decode the sample based at least in part upon an output of a combination of a deep structure and a context-dependent Hidden Markov Model (HMM), wherein the deep structure is configured to output a posterior probability of a context-dependent unit. The deep structure is a Deep Belief Network consisting of many layers of nonlinear units with connecting weights between layers trained by a pretraining step followed by a fine-tuning step.

摘要翻译： 本文公开了一种包括使处理器接收样本的动作的方法，其中样本是口语发音之一，在线手写样本或运动图像样本之一。 该方法还包括使处理器至少部分地基于深结构和上下文相关隐马尔可夫模型（HMM）的组合的输出来解码样本的动作，其中深结构被配置为输出 上下文相关单位的后验概率。 深层结构是由许多非线性单元组成的深层信念网络，其中层之间的连接权重通过预培训步骤后跟微调步骤训练。

公开(公告)号：WO2019155064A1

公开(公告)日：2019-08-15

申请号：PCT/EP2019/053322

申请日：2019-02-11

申请人： DEEPMIND TECHNOLOGIES LIMITED

发明人： MENICK, Jacob Lee ,  GRAVES, Alexander Benjamin

IPC分类号： G06N3/04 ,  G06N3/08 ,  G06N3/00 ,  G10L19/00 ,  H03M7/40

CPC分类号： G06N3/0454 ,  G06N3/006 ,  G06N3/084 ,  G06N3/088 ,  G10L19/00 ,  H03M7/40

摘要： Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an encoder neural network, a decoder neural network, and a prior neural network, and using the trained networks for generative modeling, data compression, and data decompression. In one aspect, a method comprises: providing a given observation as input to the encoder neural network to generate parameters of an encoding probability distribution; determining an updated code for the given observation; selecting a code that is assigned to an additional observation; providing the code assigned to the additional observation as input to the prior neural network to generate parameters of a prior probability distribution; sampling latent variables from the encoding probability distribution; providing the latent variables as input to the decoder neural network to generate parameters of an observation probability distribution; and determining gradients of a loss function.

公开(公告)号：WO2019126755A1

公开(公告)日：2019-06-27

申请号：PCT/US2018/067298

申请日：2018-12-21

申请人： FUGRO N.V. ,  MARTIN, Dickie Andrew ,  GAUDET, Chase John

发明人： MARTIN, Dickie Andrew ,  GAUDET, Chase John

IPC分类号： G06N20/00

CPC分类号： G06N3/0445 ,  G05D1/00 ,  G06N3/0454 ,  G06N3/084 ,  G06N3/086

摘要： Disclosed are systems, methods and computer readable mediums for generating and applying training data to training data computer learning systems and for autonomous operation of surface and sub-surface vessels. The systems, methods and computer readable mediums can include a data source, the data source providing one or more of vehicle data, operational data, and environmental data and can further include a processor, the processor analyzing at least data from the data source and generating one or more control outputs based at least in part on the analysis.