REDES NEURONALES

VolverVolver

Resultados 104 resultados LastUpdate Última actualización 03/10/2022 [14:18:00] pdf PDF xls XLS

Solicitudes publicadas en los últimos 30 días / Applications published in the last 30 days



Página1 de 5 nextPage   por página


SUBMARINE CABLE FAULT DIAGNOSIS METHOD AND APPARATUS, AND DEVICE

NºPublicación: WO2022198899A1 29/09/2022

Solicitante:

CANGZHOU POWER SUPPLY COMPANY STATE GRID HEBEI ELECTRIC POWER CO LTD [CN]
STATE GRID HEBEI ELECTRIC POWER CO LTD [CN]

CN_113298110_A

Resumen de: WO2022198899A1

The present invention is suitable for the technical field of submarine cable maintenance, and provides a submarine cable fault diagnosis method and apparatus, and a device. The method comprises: obtaining deployment data and sensing data of a target submarine cable; and inputting the deployment data and the sensing data into a trained fault diagnosis model to obtain the diagnosis result of the target submarine cable, the fault diagnosis model being a probabilistic neural network model, and the diagnosis result comprising a fault type. In the present invention, a trained probabilistic neural network model is taken as the fault diagnosis model to perform fault diagnosis on the target submarine cable, such that the fault type of the target submarine cable can be rapidly and accurately obtained.

traducir

APPARATUS AND METHOD FOR LOW-MEMORY RESIDUAL LEARNING

NºPublicación: WO2022199811A1 29/09/2022

Solicitante:

HUAWEI TECH CO LTD [CN]
MARRAS IOANNIS [DE]

Resumen de: WO2022199811A1

An apparatus (2000, 2101) for processing an input signal comprising an input tensor (401, 501, 551, 601, 801), the apparatus having a processor (2001, 2104) and being configured to implement a convolutional neural network (400, 500, 550, 600, 800) for performing residual learning, the network comprising one or more convolutional layers (402, 502, 504, 505, 552, 553, 554, 602, 802), each layer comprising a plurality of convolutional filters (403, 404, 603, 604, 803, 804), wherein, for at least one of the layers, at least some of the convolutional filters (404, 604, 804) of the respective layer (402, 602, 802) are configured to: propagate a representation (406, 501, 556, 601, 806) of the input tensor through the respective layer of the network; and append the propagated representation of the input tensor to features (405, 506, 557, 605, 805) derived from other convolutional filters (403, 604, 803) in the respective layer to form a residual connection. The approach can be implemented in low-memory devices such as smartphones and can conveniently support all skip connection types, which may be performed without explicitly adding a skip connection.

traducir

RECURRENT NEURAL NETWORK-TRANSDUCER MODEL FOR PERFORMING SPEECH RECOGNITION

NºPublicación: WO2022203701A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

US_2022310071_PA

Resumen de: WO2022203701A1

A RNN-T model (200) includes a prediction network (300) configured to, at each time step subsequent to an initial time step, receive a sequence of non-blank symbols (301). For each non-blank symbol the prediction network is also configured to generate, using a shared embedding matrix (304), an embedding (306) of the corresponding nonblank symbol, assign a respective position vector (308) to the non-blank symbol, and weight the embedding proportional to a similarity between the embedding and the respective position vector. The prediction network is also configured to generate a single embedding vector (350) at the corresponding time step. The RNN-T model also includes a joint network (230) configured to, at each of the plurality of time steps subsequent to the initial time step, receive the single embedding vector generated as output from the prediction network at the corresponding time step and generate a probability distribution over possible speech recognition hypotheses.

traducir

MIXTURE MODEL ATTENTION FOR FLEXIBLE STREAMING AND NON-STREAMING AUTOMATIC SPEECH RECOGNITION

NºPublicación: WO2022203733A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

US_2022310073_PA

Resumen de: WO2022203733A1

A method (500) for an automated speech recognition (ASR) model (200) for unifying streaming and non-streaming speech recognition including receiving a sequence of acoustic frames (110). The method includes generating, using an audio encoder (300) of an automatic speech recognition (ASR) model, a higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method further includes generating, using a joint network (230) of the ASR model, a probability distribution over possible speech recognition hypothesis at the corresponding time step based on the higher order feature representation generated by the audio encoder at the corresponding time step. The audio encoder includes a neural network that applies mixture model (MiMo) attention to compute an attention probability distribution function (PDF) using a set of mixture components of softmaxes over a context window.

traducir

GENERATING CONFIDENCE-ADAPTIVE PIXEL-LEVEL PREDICTIONS UTILIZING A MULTI-EXIT PIXEL-LEVEL PREDICTION NEURAL NETWORK

NºPublicación: US2022309285A1 29/09/2022

Solicitante:

ADOBE INC [US]

Resumen de: US2022309285A1

The present disclosure relates to systems, methods, and non-transitory computer readable media for efficiently, quickly, and flexibly generating and providing pixel-wise classification predictions utilizing early exit heads of a multi-exit pixel-level prediction neural network. For example, the disclosed systems utilize a multi-exit pixel-level prediction neural network to generate classification predictions for a digital image on the pixel level. The multi-exit pixel-level prediction neural network includes a specialized architecture with early exit heads having unique encoder-decoder architectures for generating pixel-wise classification predictions at different early exit stages. In some embodiments, the disclosed systems implement a spatial confidence-adaptive scheme to mask certain predicted pixels to prevent further processing of the masked pixels and thereby reduce computation.

traducir

FEATURE DICTIONARY FOR BANDWIDTH ENHANCEMENT

NºPublicación: US2022309291A1 29/09/2022

Solicitante:

MICRON TECHNOLOGY INC [US]

KR_20220035932_PA

Resumen de: US2022309291A1

A system having multiple devices that can host different versions of an artificial neural network (ANN) as well as different versions of a feature dictionary. In the system, encoded inputs for the ANN can be decoded by the feature dictionary, which allows for encoded input to be sent to a master version of the ANN over a network instead of an original version of the input which usually includes more data than the encoded input. Thus, by using the feature dictionary for training of a master ANN there can be reduction of data transmission.

traducir

LEARNING SYSTEM FOR CONVOLUTIONAL NEURAL NETWORKS TO IMPROVE ACCURACY OF OBJECT DETECTION IN NEW ENVIRONMENTS

NºPublicación: US2022309271A1 29/09/2022

Solicitante:

ANALOG DEVICES INT UNLIMITED COMPANY [IE]

Resumen de: US2022309271A1

A method for fine-tuning a convolutional neural network (CNN) and a sensor system based on a CNN are disclosed. The sensor system may be deployed at a deployment location. The CNN may be fine-tuned for the deployment location using sensor data, e.g., images, captured by a sensor device of the sensor system at the deployment location. The sensor data may include objects that are not present in an initial data set used for training the CNN. The sensor data and the initial data set may be input to the CNN to train the CNN and obtain fine-tuned parameters of the CNN. The CNN can thus be fine-tuned to the deployment location of the sensor system, with an increased chance of recognizing objects when using the sensor system and the CNN to recognize objects in captured sensor data.

traducir

INTELLIGENT LISTING CREATION FOR A FOR SALE OBJECT

NºPublicación: US2022309562A1 29/09/2022

Solicitante:

MERCARI INC [US]

US_2021406988_A1

Resumen de: US2022309562A1

Disclosed herein are embodiments for intelligent listing creation for a for sale object (FSO). Some embodiments operate by determining a numerical identifier corresponding to a category of the FSO. A binarization of the numerical identifier using hot encoding is performed and using a neural network regression model, an optimal offer price is generated based on a category of the FSO. Information about the FSO is provided to the neural network regression model that tokenizes the textual input, and a unique binary vector representing the category is provided instead of the numerical identifier to the neural network regression model. An optimal price, generated by the neural network regression model, based on the unique binary vector representing the category.

traducir

METHODS AND DEVICES FOR IRREGULAR PRUNING FOR AUTOMATIC SPEECH RECOGNITION

NºPublicación: US2022310069A1 29/09/2022

Solicitante:

KWAI INC [US]

Resumen de: US2022310069A1

A method and an apparatus for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a pre-determined percentage of weights from at least one block for block-wise pruning; and generating a block-wise pruned weight matrix by setting the pre-determined percentage of weights selected from the at least one block to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors.

traducir

PHYSICAL ENVIRONMENT INTERACTION WITH AN EQUIVARIANT POLICY

NºPublicación: US2022309773A1 29/09/2022

Solicitante:

BOSCH GMBH ROBERT [DE]
KONINKLIJKE PHILIPS NV [NL]

KR_20220058916_PA

Resumen de: US2022309773A1

Some embodiments are directed to a computer-implemented method of interacting with a physical environment according to a policy. The policy determines multiple action probabilities of respective actions based on an observable state of the physical environment. The policy includes a neural network parameterized by a set of parameters. The neural network determines the action probabilities by determining a final layer input from an observable state and applying a final layer of the neural network to the final layer input. The final layer is applied by applying a linear combination of a set of equivariant base weight matrices to the final layer input. The base weight matrices are equivariant in the sense that, for a set of multiple predefined transformations of the final layer input, each transformation causes a corresponding predefined action permutation of the base weight matrix output for the final layer input.

traducir

Mixture Model Attention for Flexible Streaming and Non-Streaming Automatic Speech Recognition

NºPublicación: US2022310074A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

US_2022310073_PA

Resumen de: US2022310074A1

A method for an automated speech recognition (ASR) model for unifying streaming and non-streaming speech recognition including receiving a sequence of acoustic frames. The method includes generating, using an audio encoder of an automatic speech recognition (ASR) model, a higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method further includes generating, using a joint encoder of the ASR model, a probability distribution over possible speech recognition hypothesis at the corresponding time step based on the higher order feature representation generated by the audio encoder at the corresponding time step. The audio encoder comprises a neural network that applies mixture model (MiMo) attention to compute an attention probability distribution function (PDF) using a set of mixture components of softmaxes over a context window.

traducir

Mixture Model Attention for Flexible Streaming and Non-Streaming Automatic Speech Recognition

NºPublicación: US2022310073A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

US_2022310074_PA

Resumen de: US2022310073A1

A method for an automated speech recognition (ASR) model for unifying streaming and non-streaming speech recognition including receiving a sequence of acoustic frames. The method includes generating, using an audio encoder of an automatic speech recognition (ASR) model, a higher order feature representation for a corresponding acoustic frame in the sequence of acoustic frames. The method further includes generating, using a joint encoder of the ASR model, a probability distribution over possible speech recognition hypothesis at the corresponding time step based on the higher order feature representation generated by the audio encoder at the corresponding time step. The audio encoder comprises a neural network that applies mixture model (MiMo) attention to compute an attention probability distribution function (PDF) using a set of mixture components of softmaxes over a context window.

traducir

TWO-PASS END TO END SPEECH RECOGNITION

NºPublicación: US2022310072A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

JP_2022534888_A

Resumen de: US2022310072A1

Two-pass automatic speech recognition (ASR) models can be used to perform streaming on-device ASR to generate a text representation of an utterance captured in audio data. Various implementations include a first-pass portion of the ASR model used to generate streaming candidate recognition(s) of an utterance captured in audio data. For example, the first-pass portion can include a recurrent neural network transformer (RNN-T) decoder. Various implementations include a second-pass portion of the ASR model used to revise the streaming candidate recognition(s) of the utterance and generate a text representation of the utterance. For example, the second-pass portion can include a listen attend spell (LAS) decoder. Various implementations include a shared encoder shared between the RNN-T decoder and the LAS decoder.

traducir

Artificial Intelligence System for Capturing Context by Dilated Self-Attention

NºPublicación: US2022310070A1 29/09/2022

Solicitante:

MITSUBISHI ELECTRIC RES LABORATORIES INC [US]

WO_2022201646_A1

Resumen de: US2022310070A1

An artificial intelligence (AI) system is disclosed. The AI system includes a processor that processes a sequence of input frames with a neural network including a dilated self-attention module trained to compute a sequence of outputs by transforming each input frame into a corresponding query frame, a corresponding key frame, and a corresponding value frame leading to a sequence of key frames, a sequence of value frames, and a sequence of query frames of same ordering and by performing attention calculations for each query frame with respect to a combination of a portion of the sequences of key and value frames restricted based on a location of the query frame and a dilation sequence of the key frames and a dilation sequence of value frames extracted by processing different frames of the sequences of key and value frames with a predetermined extraction function. Further, the processor renders the sequence of outputs.

traducir

METHODS AND DEVICES FOR STRUCTURED PRUNING FOR AUTOMATIC SPEECH RECOGNITION

NºPublicación: US2022310068A1 29/09/2022

Solicitante:

KWAI INC [US]

Resumen de: US2022310068A1

Methods and apparatuses for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a set of blocks from the plurality of blocks for block-wise pruning by minimizing a cost function subject to a pre-determined block-wise constraint; and generating a block-wise pruned weight matrix by setting one or more weights in the set of blocks to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors.

traducir

Audio-Visual Separation of On-Screen Sounds Based on Machine Learning Models

NºPublicación: US2022310113A1 29/09/2022

Solicitante:

GOOGLE LLC [US]

Resumen de: US2022310113A1

Apparatus and methods related to separation of audio sources are provided. The method includes receiving an audio waveform associated with a plurality of video frames. The method includes estimating, by a neural network, one or more audio sources associated with the plurality of video frames. The method includes generating, by the neural network, one or more audio embeddings corresponding to the one or more estimated audio sources. The method includes determining, based on the audio embeddings and a video embedding, whether one or more audio sources of the one or more estimated audio sources correspond to objects in the plurality of video frames. The method includes predicting, by the neural network and based on the one or more audio embeddings and the video embedding, a version of the audio waveform comprising audio sources that correspond to objects in the plurality of video frames.

traducir

METHODS AND ELECTRONIC DEVICES FOR DETECTING OBJECTS IN SURROUNDINGS OF A SELF-DRIVING CAR

NºPublicación: US2022309794A1 29/09/2022

Solicitante:

YANDEX SELF DRIVING GROUP LLC [RU]

RU_2767831_C1

Resumen de: US2022309794A1

A method and electronic device for detecting an object are disclosed. The method includes generating a cluster of points representative of the surroundings of the SDC, generating by a first Neural Network (NN) a first feature vector based on the cluster indicative of a local context of the given object in the surroundings of the SDC, generating by a second NN second feature vectors for respective points from the cluster based on a portion of the point cloud, where a given second feature vector is indicative of the local and global context of the given object, generating by the first NN a third feature vector for the given object based on the second feature vectors representative of the given object, and generating by a third NN a bounding box around the given object using the third feature vector.

traducir

MACHINE LEARNING FOR DATA EXTRACTION

NºPublicación: US2022309813A1 29/09/2022

Solicitante:

JUMIO CORP [US]

WO_2021126229_A1

Resumen de: US2022309813A1

Computer systems and methods are provided for extracting information from an image of a document. A computer system receives image data, the image data including an image of a document. The computer system determines a portion of the received image data that corresponds to a predefined document field. The computer system utilizes a neural network system to assign a label to the determined portion of the received image data. The computer system performs text recognition on the portion of the received image data and stores the recognized text in association with the assigned label.

traducir

ACCELERATED DOCUMENT CATEGORIZATION USING MACHINE LEARNING

NºPublicación: US2022309089A1 29/09/2022

Solicitante:

SARTORIUS STEDIM DATA ANALYTICS AB [SE]

WO_2022200146_A1

Resumen de: US2022309089A1

A computer-implemented method is provided. The method may comprise: obtaining at least one document to be classified; classifying, using a machine learning model including an artificial neural network (ANN) and an attention mechanism, the at least one document into at least two classes; determining, for each of the at least one document, a confidence value of the classifying, based on one or more outputs of one or more nodes comprised in the ANN; assigning, to each of the at least one document, based at least in part on the confidence value, one of at least two categories that are associated with different degrees of credibility of the classifying; and providing for display one or more of the at least one document with: the assigned category and attention information that indicates significance of one or more parts of each document provided for display in the classifying of said document.

traducir

IDENTIFYING DIGITAL ATTRIBUTES FROM MULTIPLE ATTRIBUTE GROUPS UTILIZING A DEEP COGNITIVE ATTRIBUTION NEURAL NETWORK

NºPublicación: US2022309093A1 29/09/2022

Solicitante:

ADOBE INC [US]

US_2021073267_A1

Resumen de: US2022309093A1

The present disclosure relates to systems, methods, and non-transitory computer-readable media for generating tags for an object portrayed in a digital image based on predicted attributes of the object. For example, the disclosed systems can utilize interleaved neural network layers of alternating inception layers and dilated convolution layers to generate a localization feature vector. Based on the localization feature vector, the disclosed systems can generate attribute localization feature embeddings, for example, using some pooling layer such as a global average pooling layer. The disclosed systems can then apply the attribute localization feature embeddings to corresponding attribute group classifiers to generate tags based on predicted attributes. In particular, attribute group classifiers can predict attributes as associated with a query image (e.g., based on a scoring comparison with other potential attributes of an attribute group). Based on the generated tags, the disclosed systems can respond to tag queries and search queries.

traducir

UTILIZING A NEURAL NETWORK MODEL TO GENERATE A REFERENCE IMAGE BASED ON A COMBINATION OF IMAGES

NºPublicación: US2022309631A1 29/09/2022

Solicitante:

ACCENTURE GLOBAL SOLUTIONS LTD [IE]

EP_4064078_PA

Resumen de: US2022309631A1

A device may receive complex data from a user device and may provide multiple images to the user device based on receiving the complex data. The device may receive, from the user device, a selection of two or more images from the multiple images, and may determine whether a combination of the two or more images is stored in a data structure. The device may determine a mapping of information identifying the two or more images with the complex data, based on the combination of the two or more images not being stored in the data structure, and may store the information identifying the two or more images, the complex data, and the mapping in the data structure. The device may process the two or more images to generate a reference image that satisfies a memorability score threshold and may provide the reference image to another user device.

traducir

QUANTIZATION METHOD, QUANTIZATION DEVICE, AND RECORDING MEDIUM

NºPublicación: US2022309321A1 29/09/2022

Solicitante:

PANASONIC IP MAN CO LTD [JP]

Resumen de: US2022309321A1

A quantization method executed by a computer includes: searching for quantization step sizes of parameters of a target layer by using a second inference contribution degree and quantization errors before and after quantization of the parameters of the target layer, the second inference contribution degree indicating a degree of influence of a layer next to the target layer and being obtained using a first inference contribution degree calculated in advance, the layer next to the target layer including second neurons as elements, and the first inference contribution degree indicating a degree of influence of each of layers that constitute a model composed of a neural network and each include first neurons as elements on an inference result obtained by using the model; and quantizing the parameters by using the quantization step sizes obtained as a result of the searching.

traducir

ERROR COMPENSATION IN ANALOG NEURAL NETWORKS

NºPublicación: US2022309331A1 29/09/2022

Solicitante:

AMS INT AG [CH]

DE_112020003050_T5

Resumen de: US2022309331A1

A computer-implemented method for compensation of errors due to fabrication tolerance in an analog neural network is described. The method includes: receiving a set of input weights from a trained digital neural network, the digital neural network having the same architecture as the analog neural network and being trained in a digital environment without errors due to fabrication tolerance; loading the set of input weights to the analog neural network; receiving (i) a set of test inputs for error compensation, and (ii) a set of expected outputs that is obtained by processing the set of test inputs using the trained digital neural network; processing the set of test inputs using the analog neural network to generate a set of test outputs; processing the set of test outputs and the set of expected outputs to generate a set of updated weights for the analog neural network; and loading the set of updated weights to the analog neural network.

traducir

GRAPH NEURAL NETWORKS FOR DATASETS WITH HETEROPHILY

NºPublicación: US2022309334A1 29/09/2022

Solicitante:

ADOBE INC [US]

CN_115130643_PA

Resumen de: US2022309334A1

Techniques are provided for training graph neural networks with heterophily datasets and generating predictions for such datasets with heterophily. A computing device receives a dataset including a graph data structure and processes the dataset using a graph neural network. The graph neural network defines prior belief vectors respectively corresponding to nodes of the graph data structure, executes a compatibility-guided propagation from the set of prior belief vectors and using a compatibility matrix. The graph neural network predicts predicting a class label for a node of the graph data structure based on the compatibility-guided propagations and a characteristic of at least one node within a neighborhood of the node. The computing device outputs the graph data structure where it is usable by a software tool for modifying an operation of a computing environment.

traducir

INFORMATION PROCESSING APPARATUS, VEHICLE, AND STORAGE MEDIUM

Nº publicación: US2022309797A1 29/09/2022

Solicitante:

HONDA MOTOR CO LTD [JP]

CN_115129767_PA

Resumen de: US2022309797A1

An information processing apparatus includes a processing unit configured to execute an inference operation in an execution cycle. The inference operation is executed by inputting input data including time series data to a neural network. An interval of acquiring constituting data of the time series data to be input in a single time of the inference operation is longer than the execution cycle.

traducir

Página1 de 5 nextPage por página

punteroimgVolver