Abstract: A side information calculating unit (110) calculates side information for assisting either identification processing or classification processing. When there is a discrepancy between a processing result of either the identification processing or the classification processing, and the side information, the multilayer neural network (120) changes an output value of an intermediate layer (20) and performs either the identification processing or the classification processing again.
Abstract: Aspects of the present disclosure describe systems, methods and structures providing wide-area traffic monitoring based on distributed fiber-optic sensing (DFOS) that employs deep neural network(s) for denoising noisy waterfall traces measured by the DFOS. Such systems, methods, and structures according to aspects of the present disclosure may advantageously monitor multiple highways/roadways using a single interrogator and optical fiber switch(es) which provides traffic information along every sensing point of existing, deployed, in-service optical telecommunications facilities.
Abstract: An image processing method includes an acquisition step of acquiring image data by capturing printed matter, wherein first information and second information, indicating a type of the first information, are embedded in the printed matter as an electronic watermark, an extraction step of extracting the first information and the second information from the image data acquired at the acquisition step, and a processing step of processing the extracted first information by different processing methods in accordance with the extracted second information. Based on that the type of the first information indicated by the second information is a predetermined type, the extracted first information is processed at the processing step, by a predetermined processing method for access to an external device using the extracted first information and display of a web page based on the access.
Abstract: A system with a multiplication circuit having a plurality of multipliers is disclosed. Each of the plurality of multipliers is configured to receive a data value and a weight value to generate a product value in a convolution operation of a machine learning application. The system also includes an accumulator configured to receive the product value from each of the plurality of multipliers and a register bank configured to store an output of the convolution operation. The accumulator is further configured to receive a portion of values stored in the register bank and combine the received portion of values with the product values to generate combined values. The register bank is further configured to replace the portion of values with the combined values.
Abstract: Systems and methods for implementing a spatial and temporal attention-based gated recurrent unit (GRU) for node classification over temporal attributed graphs are provided. The method includes computing, using a GRU, embeddings of nodes at different snapshots. The method includes performing weighted sum pooling of neighborhood nodes for each node. The method further includes concatenating feature vectors for each node. Final temporal network embedding vectors are generated based on the feature vectors for each node. The method also includes applying a classification model based on the final temporal network embedding vectors to the plurality of nodes to determine temporal attributed graphs with classified nodes.
Abstract: Examples disclosed herein relate to determining a response based on hierarchical models. In one implementation, a processor applies a first model to an image of an environment to select a second model. The processor applies the selected second model to the image and creates an environmental description representation based on the output of the second model. The processor determines a response based on the environmental description information.
Type:
Grant
Filed:
February 21, 2018
Date of Patent:
September 27, 2022
Assignee:
Hewlett-Packard Development Company, L.P.
Abstract: An autonomic function is caused to execute in an artificial intelligence environment to detect a new problem space. Using the autonomic function, a first model is selected. The first model includes a first trained neural network corresponding to a first ontology. A second model is automatically identified. the second model includes a second trained neural network corresponding to a second ontology. A layer is autonomically extracted from the second model and inserted into a location in the first model. A vector transformation is automatically constructed to transform an output vector of a previous layer in an immediately previous location in the model relative to the location. The layer is automatically fused in the first model using the transformed output vector as input to the layer, the fusing forming a fused model that is operable on an ontology of the new problem space.
Type:
Grant
Filed:
November 15, 2019
Date of Patent:
September 27, 2022
Assignee:
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventors:
Aaron K. Baughman, Michael Behrendt, Shikhar Kwatra, Craig M. Trim
Abstract: Methods and systems for collecting high definition images of spent firearm cartridges under different illumination conditions described herein. Features indicative of firing pin impact with each spent firearm cartridge are extracted and compared to features extracted from different spent firearm cartridges. The likelihood that the cartridges were fired from the same firearm is determined based on the differences between the extracted features. A cartridge fixture locates a spent firearm cartridge inside an imaging chamber illuminated by different combinations of illumination devices located in different locations with respect to the spent firearm cartridge. Collected images are filtered by a trained image feature filter to extract features indicative of a firing pin strike. Features extracted from different spent firearm cartridges are compared to determine the likelihood that the spent firearm cartridges were fired from the same firearm based on one or more error metrics characterizing feature differences.
Abstract: A method for ensuring safety of humans within operating area or in close proximity to an automatic apparatus is applied in and by a control apparatus. The control apparatus is coupled to one or more cameras arranged around the operating area of the automatic apparatus. The control apparatus uses deep learning techniques to analyze images captured by the cameras to determine whether there is a person in the operating area and powers off the automatic apparatus if any person is deemed present.
Type:
Grant
Filed:
March 3, 2020
Date of Patent:
September 27, 2022
Assignee:
Shenzhen Fugui Precision Ind. Co., Ltd.
Inventors:
Chang-Ching Liao, Shao-Wen Wang, Shih-Cheng Wang
Abstract: A method and system for classifying image features using a neural network is provided. The method includes training the neural network using triplet loss processes including receiving an anchor image, selecting a positive image and a negative image, generating a image embedding associated with each of the anchor image, the positive image, and the negative image, classifying image features extracted from the anchor image based on the image embedding of the anchor image, determining an image label location associated with the classified image features, extracting features associated with the determined image label location, and classifying the features associated with the determined image label location; and combining the multi-label loss with localized image classification loss and the triplet loss using a weighted loss sum.
Abstract: A method and apparatus for depth estimation of a monocular image, and a storage medium are provided. The method includes: obtaining, through a depth estimation neural network, a global feature of a monocular image according to absolute features of preset regions and relative features among the preset regions in the monocular image; and obtaining a predicted depth map of the monocular image according to the global feature, and the absolute features of preset regions and relative features among the preset regions in the monocular image.
Type:
Grant
Filed:
March 26, 2020
Date of Patent:
September 13, 2022
Assignee:
SHENZHEN SENSETIME TECHNOLOGY CO., LTD.
Inventors:
Yukang Gan, Xiangyu Xu, Wenxiu Sun, Liang Lin
Abstract: This application relates to a feature information extraction method and apparatus, a server cluster, and a storage medium. In various implementations, package attribute vectors respectively corresponding to at least two virtual item packages of a target object may be obtained. Feature extraction may be performed on these package attribute vectors to obtain feature vectors. Using the feature vector feature information may be obtained for the virtual item packages. In this way, differences between users of different attributes when the users are using virtual item packages may be considered, thereby improving the accuracy, efficiency and security of feature information extraction.
Type:
Grant
Filed:
June 12, 2019
Date of Patent:
September 6, 2022
Assignee:
Tencent Technology (Shenzhen) Company Limited
Abstract: The present disclosure relates generally to machine learning. More particularly, the present disclosure relates to systems and methods that regularize neural networks by decorrelating neurons or other parameters of the neural networks during training of the neural networks promoting these parameter to innovate over one another.
Abstract: The present invention relates to systems and methods of express tracking for patient flow management in a distributed environment. Particularly, aspects are directed to a computer implemented method that includes initiating a check-in process that includes prompting a user to scan an identifier, processing the identifier using a cascade machine-learning architecture comprising of a multi-task convolutional neural network model and a recurrent neural network model to obtain the classification of the identifier and member identification information, determining whether the user is known user based on the member identification information, when the user is a known user, verifying user data saved in the computing system associated with the identifier, once the user data is verified, determining whether the user has a scheduled appointment; and when the user has a scheduled appointment, checking the user in for the scheduled appointment.
Type:
Grant
Filed:
October 30, 2019
Date of Patent:
August 30, 2022
Assignee:
Laboratory Corporation of America Holdings
Abstract: According to an embodiment, a document analysis device includes one or more hardware processors configured to function as a sentence extraction unit, an analysis unit, a neural network unit. The analysis unit generates, for each of sentences, initial element information representing an initial value of relevance to each of predetermined attribute items. The neural network unit receives sentence information and outputs execution result of a main task on the target document, for each of the sentences. The neural network unit includes an attention unit and a main task execution. The attention receives the sentence information and the initial element information, calculates an attention weight and outputs attention information according to the attention weight, for each of the sentences. The main task execution unit executes the main task based on the attention information for each of the sentences.
Abstract: Disclosed herein is a method of facilitating charging of portable power sources using a charging system, in accordance with some embodiments. Accordingly, the method may include a step of receiving, using a communication interface, a power source information corresponding to a power source attribute from a user device, a step of analyzing, using a processing device, the power source information, a step of identifying, using the processing device, a portable power source of the portable power sources based the analyzing, a step of retrieving, using a memory device, a standard power source information corresponding to the power source attribute of the portable power source based on the identifying, a step of comparing, using the processing device, the power source information and the standard power source information, and a step of generating, using the processing device, a charging command based on the comparing.
Abstract: A technology that can enhance the computing performance of a computing system using reservoir computing (RC), includes a computing system which performs computation using a recurrent neural network (RNN) including an input unit, a reservoir unit, and an output unit. The reservoir unit includes a plurality of nodes circularly connected to each other. The circular connection has a weight matrix for determining a weight between the nodes of the plurality of nodes, in which a weight between the nodes closely arranged on the circle is larger than a weight between the nodes arranged away from each other on the circle. The plurality of nodes each have a g value that is a parameter for designating nonlinearity of an activation function of each of the nodes, and that is set so as to periodically change in a direction on the circle.
Abstract: Aspects of the disclosure provide methods and apparatuses for point cloud compression and decompression. In some examples, an apparatus for point cloud compression/decompression includes processing circuitry. For example, the apparatus is for point cloud decompression. The processing circuitry decodes prediction information of an image from a coded bitstream corresponding to a point cloud. The prediction information indicates that the image includes a plurality of missed points from at least a patch for the point cloud, and the plurality of missed points are arranged in the image according to a non-jumpy scan. Then, the processing circuitry reconstructs the plurality of missed points from the image according to the non-jumpy scan.
Abstract: In a general aspect, medical images are segmented by a medical image segmentation system. In some aspects, a medical image segmentation method includes obtaining a medical image comprising a series of images; obtaining a surrogate context based on a support set of images; identifying a query set of images; providing the surrogate context, the support set of images and the query set of images as inputs to a machine learning process; and segmenting the query set by operation of the machine learning process executed on a computer system, wherein executing the machine learning process produces segmentation information of the query set. The support set of images includes a first subset of the series of images and the query set of images includes a second subset of the series of images.
Abstract: Systems and methods for anomaly detection are provided. The method includes structuring a multi-channel spatial-temporal sequence as a four-dimensional array. The method also includes decomposing the four-dimensional array to form a low-rank component representing a background signal and a residual component representing anomalies for each time point of the multi-channel spatial-temporal sequence. The method further includes determining a sequence of anomaly maps by stacking the residual components at all time points together. Anomalies are identified based on the sequence of anomaly maps.