45th week of 2021 patent applcation highlights part 47 |
Patent application number | Title | Published |
20210350145 | OBJECT RECOGNITION METHOD OF AUTONOMOUS DRIVING DEVICE, AND AUTONOMOUS DRIVING DEVICE - Disclosed is an object recognition method including: obtaining a first RGB image by using a camera; predicting at least one first region, in which an object is unrecognizable, in the first RGB image based on brightness information of the first RGB image; determining at least one second region, in which an object exists, from among the at least one first region, based on object information obtained through a dynamic vision sensor; obtaining an enhanced second RGB image by controlling photographic configuration information of the camera in relation to the at least one second region; and recognizing the object in the second RGB image. | 2021-11-11 |
20210350146 | Vehicle Tracking Method, Apparatus, and Electronic Device - A vehicle tracking method, apparatus, and electronic device relate to the technical field of computer vision and deep learning. A method includes identifying first position information of a first vehicle in a first image of a video stream collected during driving of vehicles; and identifying second position information of a second vehicle in a second image of the video stream. The first image is the previous N frame images adjacent to the second image in the video stream, and N is a positive integer. The method also includes predicting first position offset information of the second vehicle relative to the first vehicle on the basis of the first image and the second image; and determining a tracking result of the second vehicle on the basis of the first position information, the second position information and the first position offset information. | 2021-11-11 |
20210350147 | A MAP PARTITION SYSTEM FOR AUTONOMOUS VEHICLES - In one embodiment, a system identifies a road to be navigated by an ADV, the road being captured by one or more point clouds from one or more LIDAR sensors. The system extracts road marking information of the identified road from the point clouds, the road marking information describing one or more road markings of the identified road. The system partitions the road into one or more road partitions based on the road markings. The system generates a point cloud map based on the road partitions, where the point cloud map is utilized to perceive a driving environment surrounding the ADV. | 2021-11-11 |
20210350148 | DEVICE FOR DETERMINING LANE TYPE AND METHOD THEREOF - A device for determining lane type and method thereof are provided. The device for determining lane type according to an embodiment of the present disclosure includes a camera for acquiring an around view image around a vehicle, a GPS receiver for receiving GPS information, and a controller communicatively connected to the camera and the GPS receiver. Here, the controller is configured to recognize a scene of the image acquired by the camera, detect lanes and road markings from the recognized scene, comprise a classifier adapted based on the GPS information, classify the detected lanes and road markings by the classifier, and confirm a type of the classified lanes. | 2021-11-11 |
20210350149 | LANE DETECTION METHOD AND APPARATUS,LANE DETECTION DEVICE,AND MOVABLE PLATFORM - A lane detection method includes obtaining visual detection data via a vision sensor disposed at a movable platform, performing lane line analysis and processing based on the visual detection data to obtain lane line parameters, obtaining radar detection data via a radar sensor disposed at the movable platform, performing boundary line analysis and processing based on the radar detection data to obtain boundary line parameters, and performing data fusion according to the lane line parameters and the boundary line parameters to obtain lane detection parameters. | 2021-11-11 |
20210350150 | OBJECT DETECTION USING PLANAR HOMOGRAPHY AND SELF-SUPERVISED SCENE STRUCTURE UNDERSTANDING - In various examples, a single camera is used to capture two images of a scene from different locations. A trained neural network, taking the two images as inputs, outputs a scene structure map that indicates a ratio of height and depth values for pixel locations associated with the images. This ratio may indicate the presence of an object above a surface (e.g., road surface) within the scene. Object detection then can be performed on non-zero values or regions within the scene structure map. | 2021-11-11 |
20210350151 | METHOD FOR DETERMINING A TYPE OF PARKING SPACE - A method for determining a type of parking space for a motor vehicle includes detecting a target, evaluating an orientation of the target relative to a road, and determining a type of parking space on the basis of the evaluated orientation. The detecting the target is implemented by a frontal camera of the vehicle. | 2021-11-11 |
20210350152 | STRUCTURAL OBJECT DETECTOR FOR HIERARCHICAL ONTOLOGY FOR TRAFFIC LIGHT HANDLING - Systems and methods are provided for developing/leveraging a hierarchical ontology in traffic light perception. A hierarchical ontology representative of various traffic light characteristic (e.g., states, transitions, colors, shapes, etc.) allow for structured and/or automated annotation (in supervised machine learning), as well as the ability to bootstrap traffic light prediction. Further still, the use of a hierarchical ontology provides the ability to accommodate both coarse and fine-grained model prediction, as well as the ability to generate models that are applicable to different traffic light systems used, e.g., in different geographical regions and/or contexts. | 2021-11-11 |
20210350153 | METHOD AND APPARATUS FOR DETERMINING A LOCATION OF A SHARED VEHICLE PARK POSITION - An approach is provided for facilitating use of a shared vehicle. The approach includes receiving a request to return a shared vehicle. The request includes an image captured by a camera sensor of a device. The approach also includes processing the image to determine a drop-off location where the shared vehicle is being returned. The image includes one or more photo-identifiable objects that can be used to visually position the shared vehicle. The approach also includes determining whether the determined drop-off location is authorized or designated for returning of the shared vehicle. | 2021-11-11 |
20210350154 | VEHICLE OCCUPANT DETECTION SYSTEM - A vehicle occupant detection system includes a housing. A microprocessor is mounted in the housing and is programmed to detect an engine engagement status of a vehicle. A camera is mounted on the housing and is directed forward of the housing. The camera electrically coupled to the microprocessor is programmed with facial recognition software to compare facial images against images captured by the camera. A motion sensor mounted on the housing detects motion forward of the housing. The camera captures an image when the motion sensor detects motion. A first condition is defined when the motion sensor detects motion, the camera captures an image prompting a facial recognition match and the microprocessor detects the vehicle is parked. A sound emitter is mounted on the housing and is electrically coupled to the microprocessor and emit a low decibel sound when the first condition is first attained. | 2021-11-11 |
20210350155 | VEHICULAR MONITORING DEVICE - In this vehicular monitoring device, an imaging unit | 2021-11-11 |
20210350156 | VEHICULAR MONITORING DEVICE - In this vehicular monitoring device, an imaging unit | 2021-11-11 |
20210350157 | METHOD AND APPARATUS FOR DETERMINING FOOTPRINT IDENTITY USING DIMENSION REDUCTION ALGORITHM - A method of determining footprint identity using a dimension reduction algorithm according to an embodiment includes: pre-processing to process three-dimensional (3D) image data about footprints of a first person and a second person and convert the 3D image data into one-dimensional (1D) data about the footprints of the first person and the second person; calculating a distribution of cross-correlation coefficients between two pieces of 1D data about footprints of the first person (SC: same footwear correlation) and a distribution of cross-correlation coefficients between the 1D data about the footprints of the first person and the second person (DC: difference footwear correlation); and calculating a likelihood ratio based on the SC and the DC to determine the degree of correspondence between the footprints of the first person and the second person. | 2021-11-11 |
20210350158 | INFORMATION PROVIDING DEVICE, INFORMATION PROVIDING METHOD, AND STORAGE MEDIUM - An information providing device according to one aspect of the present disclosure includes: at least one memory storing a set of instructions; and at least one processor configured to execute the set of instructions to: receive a face image; determine whether a person in the face image is unsuitable for iris data acquisition based on the face image; and output information based on determining that the person is unsuitable for the iris data acquisition when the person is determined to be unsuitable for the iris data acquisition. | 2021-11-11 |
20210350159 | IMAGING DEVICE AND IMAGING SYSTEM - An imaging device is mounted on or built into a moving body, and the imaging device includes: a camera that captures an image of surroundings of the moving body; an image processing part that processes an image captured by the camera; and a post-processing part that transmits or records an image processed by the image processing part, wherein the image processing part detects personal information contained in an image captured by the camera, and performs image processing for disabling determination of the personal information. | 2021-11-11 |
20210350160 | System And Method For An Activity Based Intelligence Contextualizer - The systems and methods of the present disclosure are directed to monitoring activity in an area of interest by contextualizing activity-based intelligence data received from one or more disparate data sources. The computing system selects one or more watchboxes for analyzing the features of the received data and uses one or more bounding boxes to analyze the features of the received data according to data type. The analysis includes comparing the features of the one or more bounding boxes against one or more reference data values of feature indicators associated with an activity defined by an activity model associated with the selected watchbox. The reference data of feature indicators is used to identify and track changes in spatial, temporal, aural, spectral, and other types of characteristic data to identify variations in an activity and the significance of the variations in the activity to the area of interest being monitored. | 2021-11-11 |
20210350161 | APPARATUS AND METHOD FOR PROCESSING DETECTION BOXES - A mechanism for performing non-maximum suppression (NMS) on a plurality of detection boxes identifying potential locations for one or more objects within an image. The mechanism uses a tiling system that divides the image into a plurality of tiles. A tile-by-tile suppression process is performed, in which at least some detection boxes that overlap a particular tile are processed to determine whether any detection boxes are to be discarded. | 2021-11-11 |
20210350162 | VISUAL OBSERVER FOR UNMANNED AERIAL VEHICLES - In some examples, a device may receive, from a first camera, a plurality of images of an airspace corresponding to an area of operation of an unmanned aerial vehicle (UAV). The device may detect, based on the plurality of images from the first camera, a candidate object approaching or within the airspace. Based on detecting the candidate object, the device may control a second camera to direct a field of view of the second camera toward the candidate object. Further, based on images from the second camera captured at a first location and images from at least one other camera captured at a second location, the candidate object may be determined to be an object of interest. In addition, at least one action may be taken based on determining that the candidate object is the object of interest. | 2021-11-11 |
20210350163 | Equalization-Based Image Processing and Spatial Crosstalk Attenuator - The technology disclosed attenuates spatial crosstalk from sequencing images for base calling. In particular, the technology disclosed accesses an image whose pixels depict intensity emissions from a target cluster and intensity emissions from additional adjacent clusters. The pixels include a center pixel that contains a center of the target cluster. Each pixel in the pixels is divisible into a plurality of subpixels. Depending upon a particular subpixel, in a plurality of subpixels of the center pixel, which contains the center of the target cluster, the technology disclosed selects, from a bank of subpixel lookup tables, a subpixel lookup table that corresponds to the particular subpixel. The selected subpixel lookup table contains pixel coefficients that are configured to maximizes a signal-to-noise ratio. The technology disclosed element-wise multiplies the pixel coefficients with the pixels and determines a weighted sum. | 2021-11-11 |
20210350164 | ELECTRONIC DEVICE AND CONTROL METHOD THEREFOR - The present disclosure relates to an artificial intelligence (AI) system that utilizes a machine learning algorithm and an application thereof. Disclosed is an electronic device. The electronic device comprises: a memory in which a first filter for identifying an input image is stored; and a processor for rotating between a plurality of elements included in the memory in which the first filter is stored and a plurality of elements included in the first filter, obtaining at least one second filter by scaling a filter region including at least some of the plurality of elements, and identifying the input image on the basis of a result value obtained by performing convolution on a pixel value included in the input image with each of the first filter and the second filter. | 2021-11-11 |
20210350165 | METHODS AND APPARATUS FOR GENERATING POINT CLOUD HISTOGRAMS - The techniques described herein relate to methods, apparatus, and computer readable media configured to generate point cloud histograms. A one-dimensional histogram can be generated by determining a distance to a reference for each 3D point of a 3D point cloud. A one-dimensional histogram is generated by adding, for each histogram entry, distances that are within the entry's range of distances. A two-dimensional histogram can be determined by generating a set of orientations by determining, for each 3D point, an orientation with at least a first value for a first component and a second value for a second component. A two-dimensional histogram can be generated based on the set of orientations. Each bin can be associated with ranges of values for the first and second components. Orientations can be added for each bin that have first and second values within the first and second ranges of values, respectively, of the bin. | 2021-11-11 |
20210350166 | SYSTEMS AND METHODS TO PROCESS ELECTRONIC IMAGES TO DETERMINE SALIENT INFORMATION IN DIGITAL PATHOLOGY - Systems and methods are disclosed for identifying a diagnostic feature of a digitized pathology image, including receiving one or more digitized images of a pathology specimen, and medical metadata comprising at least one of image metadata, specimen metadata, clinical information, and/or patient information, applying a machine learning model to predict a plurality of relevant diagnostic features based on medical metadata, the machine learning model having been developed using an archive of processed images and prospective patient data, and determining at least one relevant diagnostic feature of the relevant diagnostic features for output to a display. | 2021-11-11 |
20210350167 | DATA INGESTION PLATFORM - Embodiments are directed to data ingestion over a network. Raw data and integrated data associated with a plurality of separate data sources may be provided such that the raw data includes content associated with a plurality of subjects. Categorization models may be employed to categorize the raw data based on various features, such as, format, structure, data source, variability, volume, or associated entities. Matching models may be determined based on the categorization of the of the raw data, the integrated data and the content associated with the plurality of subjects. Matching models may generate a plurality of unified facts based on the raw data and the integrated data such that each unified fact is associated with a score associated with a quality of its match with a unified schema. | 2021-11-11 |
20210350168 | IMAGE SEGMENTATION METHOD AND IMAGE PROCESSING APPARATUS - This application discloses an image segmentation method in the field of artificial intelligence. The method includes: obtaining an input image and a processing requirement; performing multi-layer feature extraction on the input image to obtain a plurality of feature maps; downsampling the plurality of feature maps to obtain a plurality of feature maps with a reference resolution, where the reference resolution is less than a resolution of the input image; fusing the plurality of feature maps with the reference resolution to obtain at least one feature map group; upsampling the feature map group by using a transformation matrix W, to obtain a target feature map group; and performing target processing on the target feature map group based on the processing requirement to obtain a target image. | 2021-11-11 |
20210350169 | IMAGE ANNOTATION METHOD AND APPARATUS, ANNOTATION PRESENTATION METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM - A computer device obtains a to-be-annotated image having a first magnification. The device obtains an annotated image from an annotated image set, the annotated image distinct from the to-be-annotated image and having a second magnification that is distinct from with the first magnification. The annotated image set includes at least one annotated image. The device matches the to-be-annotated image with the annotated image to obtain an affine transformation matrix, and generates annotation information of the to-be-annotated image according to the affine transformation matrix and the annotated image. In this way, annotations corresponding to images at different magnifications may be migrated. For example, the annotations may be migrated from the low-magnification images to the high-magnification images, thereby reducing the manual annotation amount and avoiding repeated annotations, and further improving annotation efficiency and reducing labor costs. | 2021-11-11 |
20210350170 | LOCALIZATION METHOD AND APPARATUS BASED ON SHARED MAP, ELECTRONIC DEVICE AND STORAGE MEDIUM - Related are a localization method and apparatus based on a shared map, an electronic device and a storage medium. The method includes that: from global map data, including at least one key frame, of an image collected by a first terminal, local map data associated with the key frame are extracted; a present frame in an image collected by a second terminal is acquired; and feature matching is performed on the present frame and the local map data, and a localization result for the present frame is obtained according to a matching result. With the adoption of the present disclosure, multiple moving terminals can be accurately localized to each other in the shared map. | 2021-11-11 |
20210350171 | RESOURCE SCORING AND RECOMMENDATION SYSTEM - A computing device includes a memory and processing circuitry. The memory is configured to store an organizational proximity dataset for a current user. The processing is configured to generate scores for a plurality of resources based on the organizational proximity dataset stored to the memory for the current user. The processing circuitry is further configured to recommend one or more resources of the plurality of resources to the current user based on the scores generated for the plurality of resources. | 2021-11-11 |
20210350172 | POINT-SET KERNEL CLUSTERING - A computer-implemented clustering method is disclosed for image segmentation, social network analysis, computational biology, market research, search engine and other applications. At the heart of the method is a point-set kernel that measures the similarity between a data point and a set of data points. The method has a procedure that employs the point-set kernel to expand from a seed point to a cluster; and finally identifies all clusters in the given dataset. Applying the method for image segmentation, it identifies several segments in the image, where points in each segment have high similarity: but points in one segment have low similarity with respect to other segments. The method is both effective and efficient that enables it to deal with large scale datasets. In contrast, existing clustering methods are either efficient or effective; and even efficient ones have difficulty dealing with large scale datasets without massive parallelization. | 2021-11-11 |
20210350173 | METHOD AND APPARATUS FOR EVALUATING IMAGE RELATIVE DEFINITION, DEVICE AND MEDIUM - Provided are a method and apparatus for evaluating image relative definition, a device and a medium, relating to technologies such as computer vision, deep learning and intelligent medical. A specific implementation solution is: extracting a multi-scale feature of each image in an image set, where the multi-scale feature is used for representing definition features of objects having different sizes in an image; and scoring relative definition of each image in the image set according to the multi-scale feature by using a relative definition scoring model pre-trained, where the purpose for training the relative definition scoring model is to learn a feature related to image definition in the multi-scale feature. | 2021-11-11 |
20210350174 | Color Image Analysis for Makeup Color Prediction Model - A system generates a prediction model for makeup color matching. The system includes a data storage device and a modeling engine. The data storage device stores a plurality of color sets comprising a skin color set, a makeup color set, and a target color set. The modeling engine is coupled to the data storage device and configured to generate a prediction model and an output color set. The prediction model models generation of the target color set based on inputs from the skin color set and the makeup color set. The output color set approximates the target color set with a predictable accuracy. | 2021-11-11 |
20210350175 | KEY-VALUE MEMORY NETWORK FOR PREDICTING TIME-SERIES METRICS OF TARGET ENTITIES - This disclosure involves using key-value memory networks to predict time-series data. For instance, a computing system retrieves, for a target entity, static feature data and target time-series feature data. The computing system can normalize the target time-series feature data based on a normalization scale. The computing system also generates input data by, for example, concatenating the static feature data, the normalized time-series feature data, and time-specific feature data. The computing system generates predicted time-series data for the target metric of the target entity by applying a key-value memory network to the input data. The key-value memory network can include a key matrix learned from training static feature data and training time-series feature data, a value matrix representing time-series trends, and an output layer with a continuous activation function for generating predicted time-series data. | 2021-11-11 |
20210350176 | MULTIPLE INSTANCE LEARNER FOR PROGNOSTIC TISSUE PATTERN IDENTIFICATION - The method includes receiving digital images of tissue samples of patients, the images having assigned a label indicating a patient-related attribute value; splitting each received image into a set of image tiles; computing a feature vector for each tile; training a Multiple-Instance-Learning program on all the tiles and respective feature vectors for computing for each of the tiles a numerical value being indicative of the predictive power of the feature vector associated with the tile in respect to the label of the tile's respective image; and outputting a report gallery including tiles sorted in accordance with their respectively computed numerical value and/or including a graphical representation of the numerical value. | 2021-11-11 |
20210350177 | NETWORK TRAINING METHOD AND DEVICE AND STORAGE MEDIUM - The present disclosure relates to a network training method and device, an image processing method and device, the method comprising: performing pixel shuffling on a first image in a training set to obtain a second image, wherein the first image is an image subjected to pixel shuffling; performing, by a feature extraction network of a neural network, feature extraction on the first image to obtain a first image feature, and performing, by a feature extraction network, feature extraction on the second image to obtain a second image feature; performing, by a recognition network of the neural network, recognition on the first image feature to obtain a recognition result of the first image; and training the neural network according to the recognition result, the first image feature and the second image feature. Embodiments of the present disclosure enable improvement of recognition precision of neural networks. | 2021-11-11 |
20210350178 | INFORMATION PROCESSING DEVICE, INFORMATION PROCESSING METHOD, AND PROGRAM - It is aimed to facilitate obtaining of a large number of pieces of data for learning that are necessary to obtain a good-quality learning result. | 2021-11-11 |
20210350179 | METHOD FOR DETECTING ADVERSE CARDIAC EVENTS | 2021-11-11 |
20210350180 | SYSTEMS AND METHODS FOR DISTRIBUTED DATA ANALYTICS - The invention provides systems and method for generating device-specific artificial neural network (ANN) models for distribution across user devices. Sample datasets are collected from devices in a particular environment or use case and include predictions by device-specific ANN models executing the user devices. The received datasets are used with existing datasets and stored ANN models to generate updated device-specific ANN models from each of the stored instances of the device ANN models based on the training data. | 2021-11-11 |
20210350181 | LABEL REDUCTION IN MAINTAINING TEST SETS - A computer-implemented method, a computer program product, and a system for reducing labeled sample quantities required to update test sets. The computer-implemented method includes inputting a portion of unlabeled production data into a base model and generating labeled output relating to the unlabeled production data. The computer-implemented method also includes inputting the labeled output into a performance predictor. The performance predictor is a meta model of the base model that is trained with another portion of the unlabeled production data, a training set used to train the base model, and a test set portioned from the training set. The computer-implemented method further includes outputting, by the performance predictor, a performance metric relating to the labeled output produced by the trained base model. The performance metric can be any metric capable of measuring the output performance of the base model. | 2021-11-11 |
20210350182 | TRAINING A FUNCTION TO RESPOND PREDICTABLY TO DIFFERENCES - A computer-implemented method of training a machine learnable function, such as an image classifier or image feature extractor. When applying such machine learnable functions in autonomous driving and similar application areas, generalizability may be important. To improve generalizability, the machine learnable function is rewarded for responding predictably at a layer of the machine learnable function to a set of differences between input observations. This is done by means of a regularization objective included in the objective function used to train the machine learnable function. The regularization objective rewards a mutual statistical dependence between representations of input observations at the given layer, given a difference label indicating a difference between the input observations. | 2021-11-11 |
20210350183 | POINT CLOUD SEGMENTATION METHOD, COMPUTER-READABLE STORAGE MEDIUM, AND COMPUTER DEVICE - This application relates to a point cloud segmentation method, a computer-readable storage medium, and a computer device. The method includes encoding a to-be-processed point cloud to obtain a shared feature, the shared feature referring to a feature shared at a semantic level and at an instance level; decoding the shared feature to obtain a semantic feature and an instance feature respectively; adapting the semantic feature to an instance feature space and fusing the semantic feature with the instance feature, to obtain a semantic-fused instance feature of the point cloud, the semantic-fused instance feature representing an instance feature fused with the semantic feature; dividing the semantic-fused instance feature of the point cloud, to obtain a semantic-fused instance feature of each point in the point cloud; and determining an instance category to which each point belongs according to the semantic-fused instance feature of each point. | 2021-11-11 |
20210350184 | VISUAL BEHAVIOR GUIDED OBJECT DETECTION - A training system for a deep neural network and method of training is disclosed. The system and/or method may comprise: receiving, from an eye-tracking system associated with a sensor, an image frame captured while an operator is controlling a vehicle; receiving, from the eye-tracking system, eyeball gaze data corresponding to the image frame; and iteratively training the deep neural network to determine an object of interest depicted within the image frame based on the eyeball gaze data. The deep neural network generates at least one feature map and determine a proposed region corresponding to the object of interest within the at least one feature map based on the eyeball gaze data. | 2021-11-11 |
20210350185 | SYSTEM AND METHOD FOR GENERATING LARGE SIMULATION DATA SETS FOR TESTING AN AUTONOMOUS DRIVER - A system for creating synthetic data for testing an autonomous system, comprising at least one hardware processor adapted to execute a code for: using a machine learning model to compute a plurality of depth maps based on a plurality of real signals captured simultaneously from a common physical scene, each of the plurality of real signals are captured by one of a plurality of sensors, each of the plurality of computed depth maps qualifies one of the plurality of real signals; applying a point of view transformation to the plurality of real signals and the plurality of depth maps, to produce synthetic data simulating a possible signal captured from the common physical scene by a target sensor in an identified position relative to the plurality of sensors; and providing the synthetic data to at least one testing engine to test an autonomous system comprising the target sensor. | 2021-11-11 |
20210350186 | SYSTEMS AND METHODS FOR DETECTING LATERALITY OF A MEDICAL IMAGE - An x-ray image laterality detection system is provided. The x-ray image laterality detection system includes a detection computing device. The processor of the computing device is programmed to execute a neural network model for analyzing x-ray images, wherein the neural network model is trained with training x-ray images as inputs and observed laterality classes associated with the training x-ray images as outputs. The process is also programmed to receive an unclassified x-ray image, analyze the unclassified x-ray image using the neural network model, and assign a laterality class to the unclassified x-ray image. If the assigned laterality class is not target laterality, the processor is programmed to adjust the unclassified x-ray image to derive a corrected x-ray image having the target laterality and output the corrected x-ray image. If the assigned laterality class is the target laterality, the processor is programmed to output the unclassified x-ray image. | 2021-11-11 |
20210350187 | 3D IMAGE CLASSIFICATION METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM - The disclosure provides a three-dimensional (3D) image classification method and apparatus, a device, and a storage medium. The method includes: obtaining a 3D image, the 3D image including first-dimensional image information, second-dimensional image information, and third-dimensional image information; extracting a first image feature corresponding to planar image information from the 3D image; extracting a second image feature corresponding to the third-dimensional image information from the 3D image; fusing the first image feature and the second image feature, to obtain a fused image feature corresponding to the 3D image; and determining a classification result corresponding to the 3D image according to the fused image feature corresponding to the 3D image. | 2021-11-11 |
20210350188 | Pill Shape Classification using Imbalanced Data with Human-Machine Hybrid Explainable Model - A Human Machine Hybrid (HMH) pill shape classification system uses a decision tree with interpretable metrics. The disclosed approach for pill shape classification requires human intervention for determining the meta-classes and variables used. The creation of decision boundaries is accomplished with machine learning (ML) algorithms. Scatter plots are manually inspected to find candidate pairs of variables and potential meta-classes. | 2021-11-11 |
20210350189 | Bayesian Methodology for Geospatial Object/Characteristic Detection - A location of an object of interest ( | 2021-11-11 |
20210350190 | USING ATTRIBUTES FOR IDENTIYFING IMAGERY FOR SELECTION - A system includes a computing device that includes a memory configured to store instructions. The system also includes a processor to execute the instructions to perform operations that include receiving data representing an image, the image being represented in the data by a collection of visual elements. Operations also include determining whether to select the image for presentation by one or more entities using a machine learning system, the machine learning system being trained using data representing a plurality of training images and data representing one or more attributes regarding image presentation by the one or more entities. | 2021-11-11 |
20210350191 | TRAINING DATA GENERATING DEVICE, METHOD, AND PROGRAM, AND CROWD STATE RECOGNITION DEVICE, METHOD, AND PROGRAM - At least one storage stores a dictionary having information corresponding to crowd state images. At least one processor extracts rectangular regions, a size of the rectangular regions is predetermined, from the given image, and recognizes crowd states in the extracted rectangular regions based on the dictionary. The dictionary is acquired by machine learning by use of a plurality of pairs of crowd state images and training label for the crowd state image. | 2021-11-11 |
20210350192 | CONFIGURABLE ALARM SYSTEM COMPONENT - A component | 2021-11-11 |
20210350193 | ANTI-COUNTERFEITING IMAGE CODE EMBEDDED IN A DECORATIVE PATTERN OF A CERAMIC TILE AND ANTI-COUNTERFEITING METHOD THEREOF - The present disclosure relates to an anti-counterfeiting image code embedded in a decorative pattern of a ceramic tile and an anti-counterfeiting method thereof, the anti-counterfeiting image code is input into a terminal recognition software application. The anti-counterfeiting method includes steps of: (1) embedding an image code into the decorative pattern of the ceramic tile; (2) inputting the decorative pattern on a surface of the ceramic tile into an image code generating software to generate the image code that can be decoded, editing a ceramic tile parameter and ceramic tile information in the image code generating software; (3) packing the image code and inputting it into a terminal recognition software application; (4) downloading the terminal recognition software application at a mobile terminal; and (5) opening an application to initiate a code scanner and capturing a image or a pre-taught partial feature image. | 2021-11-11 |
20210350194 | TRANSACTION CARD ASSEMBLY - Provided are approaches for providing multiple user accounts in a same transaction card assembly. The transaction card assembly may include a first card including a first card first side opposite a first card second side, the first card first side including a first pair of magnetic stripes and the first card second side including a first pair of identification chips. The transaction card assembly may further include a second card coupled to the first card, the second card including a second card first side opposite a second card second side, the second card first side including a second pair of magnetic stripes, and the second card second side including a second pair of identification chips. The first and second cards are slidable relative to one another between multiple positions to selectively expose and cover each identification chip of the first and second pairs of identification chips. | 2021-11-11 |
20210350195 | ALARM SYSTEM COMPONENT WITH UNPOWERED EVENT DETECTION - A component | 2021-11-11 |
20210350196 | MODULE | 2021-11-11 |
20210350197 | ELECTRONIC TAG AND SYSTEM AND METHOD FOR SECURING ELECTRONIC TAG - A system for operating an electronic tag includes a reader and an operation device. The electronic tag is coupled to an object and stores a variable state code and a variable physical parameter code. The variable state code represents a variable state of the object. The variable physical parameter code represents a variable physical parameter associated with the object. Under a condition that the variable physical parameter is substantially equal to a specific physical parameter, the operation device makes a logical decision on whether the variable state code is currently equal to a first specific state code representing a first specific state. Under a condition that the logical decision is true, the operation device changes the variable state code to a second specific state code representing a second specific state, and assigns a specific physical parameter code representing the specific physical parameter to the variable physical parameter code. | 2021-11-11 |
20210350198 | RFID ENABLED METAL TRANSACTION CARDS - A transaction card (smartcard) having a front “continuous” (with no slit) metal layer (ML, CML) with an opening (MO) for a dual-interface transponder chip module (TCM) having a module antenna (MA) on its bond side. A magnetic shielding layer (MSL) comprising ferrite material disposed below the front face continuous metal layer. An amplifying element, booster antenna circuit (BAC) disposed under the magnetic shielding layer. A rear discontinuous metal layer (ML, DML) with a slit (S) and a metal ledge surrounding the module opening to function as a coupling frame (CF). A rear plastic layer formed of non-RF impeding material may support a magnetic stripe and security elements (signature panel and hologram). A portion of the front face continuous metal layer may protrude downward into the magnetic shielding layer and booster antenna circuit layer. The rear discontinuous metal layer may have an additional slit to regulate the activation distance. | 2021-11-11 |
20210350199 | WIRELESS COMMUNICATION DEVICE AND METHOD OF MANUFACTURING SAME - A wireless communication device is provided that includes an RFIC module in which an RFIC chip and first and second terminal electrodes are incorporated, and an antenna member including an antenna base material and antenna patterns including first and second coupling portions. The RFIC module and the antenna member are bonded to each other via an insulating first adhesive layer. Between the first terminal electrode and the first coupling portion and between the second terminal electrode and the second coupling portion, a distance t | 2021-11-11 |
20210350200 | Wireless Communications And Transducer Based Event Detection Platform - A low-cost, multi-function adhesive wireless communications and transducer platform with a form factor that unobtrusively integrates one or more transducers and one or more wireless communication devices in an adhesive product system. In an aspect, the adhesive product system integrates transducer and wireless communication components within a flexible adhesive structure in a way that not only provides a cost-effective platform for interconnecting, optimizing, and protecting the constituent components but also maintains the flexibility needed to function as an adhesive product that can be deployed seamlessly and unobtrusively into various applications and workflows, including sensing, notification, security, and object tracking applications, and asset management workflows such as manufacturing, storage, shipping, delivery, and other logistics associated with moving products and other physical objects. | 2021-11-11 |
20210350201 | DYNAMIC REGION BASED APPLICATION OPERATIONS - Techniques are disclosed for a hybrid undo/redo for text editing, where non-linear undo and redo operations are performed across dynamic regions in a document and linear undo and redo operations are performed within the dynamic regions in the document. In an example, the hybrid undo/redo may be achieved by maintaining respective region offset values for the dynamic regions created in a document by the edits made to the document. In operation, the respective region offset values associated with the dynamic regions can be used to negate or otherwise counteract the effect of edits made in the dynamic regions. | 2021-11-11 |
20210350202 | METHODS AND SYSTEMS OF AUTOMATIC CREATION OF USER PERSONAS - A computerized method for managing an artificially-intelligent platform to generate personas automatically from digital data includes the step of obtaining an analytics data set. The method includes the step of augmenting the analytics data set with additional context information provided by augmentation data, wherein the augmentation data comprises specified a set of external data sources and data models. The method includes the step of determining, with a specified machine learning algorithm, a set of behavioral insights from the augmented analytics data set. The method includes the step of automatically grouping a set of users of a web-application or web site based on their behavior, demographics, history of transactions, and psychographics. The method includes the step of generating a persona for each of the segment associated with a user of the set of user, wherein a segment is a group based on a user behavior, a user demographic, a user transactional history, a user psychographic attribute. | 2021-11-11 |
20210350203 | NEURAL ARCHITECTURE SEARCH BASED OPTIMIZED DNN MODEL GENERATION FOR EXECUTION OF TASKS IN ELECTRONIC DEVICE - Embodiments herein provide a NAS method of generating an optimized DNN model for executing a task in an electronic device. The method includes identifying the task to be executed in the electronic device. The method includes estimating a performance parameter to be achieved while executing the task. The method includes determining hardware parameters of the electronic device required to execute the task based on the performance parameter and the task, and determining optimal neural blocks from a plurality of neural blocks based on the performance parameter and the hardware parameter of the electronic device. The method includes generating the optimized DNN model for executing the task based on the optimal neural blocks, and executing the task using the optimized DNN model. | 2021-11-11 |
20210350204 | CONVOLUTIONAL NEURAL NETWORK ACCELERATOR - Disclosed embodiments relate to a convolutional neural network accelerator, comprising: arranging original weights in a computation sequence and aligning by bit to obtain a weight matrix, removing slack bits in the weight matrix, allowing essential bits in each column of the weight matrix to fill the vacancies according to the computation sequence to obtain an intermediate matrix, removing null rows in the intermediate matrix, obtain a kneading matrix, wherein each row of the kneading matrix serves as a kneading weight; obtaining positional information of the activation corresponding to each bit of the kneading weight; divides the kneading weight by bit into multiple weight segments, processing summation of the weight segments and the corresponding activations according to the positional information, and sending a processing result to an adder tree to obtain an output feature map by means of executing shift-and-add on the processing result. | 2021-11-11 |
20210350205 | Convolution Processing Method and Apparatus for Convolutional Neural Network, and Storage Medium - A convolution processing method and apparatus for a convolutional neural network, and a storage medium are provided. The method includes that: weight values in a sub convolution kernel in the convolutional neural network are classified; for each of the weight values, an indicator storing a corresponding operation to be executed on data and an address representing the weight value are generated according to a classification result of the corresponding weight value; corresponding to-be-processed data is acquired according to the address of the weight value; and a convolution operation is executed on the to-be-processed data according to the indicator to obtain a convolution result. | 2021-11-11 |
20210350206 | Highly Efficient Convolutional Neural Networks - The present disclosure provides directed to new, more efficient neural network architectures. As one example, in some implementations, the neural network architectures of the present disclosure can include a linear bottleneck layer positioned structurally prior to and/or after one or more convolutional layers, such as, for example, one or more depthwise separable convolutional layers. As another example, in some implementations, the neural network architectures of the present disclosure can include one or more inverted residual blocks where the input and output of the inverted residual block are thin bottleneck layers, while an intermediate layer is an expanded representation. For example, the expanded representation can include one or more convolutional layers, such as, for example, one or more depthwise separable convolutional layers. A residual shortcut connection can exist between the thin bottleneck layers that play a role of an input and output of the inverted residual block. | 2021-11-11 |
20210350207 | RECURRENT NEURAL NETWORKS FOR DATA ITEM GENERATION - Methods, and systems, including computer programs encoded on computer storage media for generating data items. A method includes reading a glimpse from a data item using a decoder hidden state vector of a decoder for a preceding time step, providing, as input to a encoder, the glimpse and decoder hidden state vector for the preceding time step for processing, receiving, as output from the encoder, a generated encoder hidden state vector for the time step, generating a decoder input from the generated encoder hidden state vector, providing the decoder input to the decoder for processing, receiving, as output from the decoder, a generated a decoder hidden state vector for the time step, generating a neural network output update from the decoder hidden state vector for the time step, and combining the neural network output update with a current neural network output to generate an updated neural network output. | 2021-11-11 |
20210350208 | METHOD AND DEVICE FOR PREDICTING PRODUCTION PERFORMANCE OF OIL RESERVOIR - The present disclosure provides a method and device for predicting a production performance of an oil reservoir. The method includes: determining a single-well numerical simulation data set according to geological parameters, rock and fluid parameters and construction data; performing reservoir numerical simulation based on the single-well numerical simulation data set, and determining a standard data set for oil reservoir production performance prediction; establishing a deep belief network (DBN) model for oil reservoir production performance prediction according to the standard data set; and predicting the production performance of a target well by using the DBN model to obtain a production performance prediction result of the target well. The present disclosure can be used to fast and accurately predict the production performance of an oil well in an unconventional oil reservoir. For a given block, the DBN model can be used indefinitely without the target well being put into production. | 2021-11-11 |
20210350209 | INTENT AND CONTEXT-AWARE DIALOGUE BASED VIRTUAL ASSISTANCE - In some examples, with respect to intent and context-aware dialogue based virtual assistance, an intent of an inquiry may be determined using an intent classification model. A determination may be made as to whether the determined intent matches a pre-specified intent of a plurality of pre-specified intents. Based on a determination that the determined intent does not match the pre-specified intent, a question related to the inquiry may be generated. Another intent of the inquiry may be determined by analyzing a response to the question using the intent classification model. A determination may be made as to whether the determined other intent matches another pre-specified intent of the plurality of pre-specified intents. Based on a determination that the determined other intent does not match the other pre-specified intent, a deep learning model may be utilized to predict a response to the inquiry. | 2021-11-11 |
20210350210 | METHOD AND APPARATUS FOR KEEPING STATISTICAL INFERENCE ACCURACY WITH 8-BIT WINOGRAD CONVOLUTION - A method and apparatus for keeping statistical inference accuracy with 8-bit winograd convolution. A calibration dataset and a pretrained CNN comprising 32-bit floating point weight values may be sampled to generate an input activation tensor and a weight tensor. A transformed input activation tensor may be generated by multiplying the input activation tensor and an input matrix to generate a transformed input activation tensor. A transformed weight tensor may be generated by multiplying the weight tensor and a weight matrix. A scale factor may be computed for each transformed tensor. An 8-bit CNN model including the scale factors may be generated. | 2021-11-11 |
20210350211 | DISTRIBUTED ARCHITECTURE FOR EXPLAINABLE AI MODELS - A method, and system for a distributed artificial intelligence architecture may be shown and described. An embodiment may present an exemplary distributed explainable neural network (XNN) architecture, whereby multiple XNNs may be processed in parallel in order to increase performance. The distributed architecture may include a parallel execution step which may combine parallel XNNs into an aggregate model by calculating the average (or weighted average) from the parallel models. A distributed hybrid XNN/XAI architecture may include multiple independent models which can work independently without relying on the full distributed architecture. An exemplary architecture may be useful for large datasets where the training data cannot fit in the CPU/GPU memory of a single machine. The component XNNs can be standard plain XNNs or any XNN/XAI variants such as convolutional XNNs (CNN-XNNs), predictive XNNS (PR-XNNs), and the like, together with the white-box portions of grey-box models like INNs. | 2021-11-11 |
20210350212 | ABSTRACTION LIBRARY TO ENABLE SCALABLE DISTRIBUTED MACHINE LEARNING - One embodiment provides for a non-transitory machine readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations comprising providing an interface to define a neural network using machine-learning domain specific terminology, wherein the interface enables selection of a neural network topology and abstracts low-level communication details of distributed training of the neural network. | 2021-11-11 |
20210350213 | AUTOMATED CONFIGURATION DETERMINATIONS FOR DATA CENTER DEVICES USING ARTIFICIAL INTELLIGENCE TECHNIQUES - Methods, apparatus, and processor-readable storage media for automated configuration determinations for data center devices using artificial intelligence are provided herein. An example computer-implemented method includes obtaining input information pertaining to one or more device-related changes to a data center; obtaining telemetry data attributed to one or more devices in the data center; determining one or more device configurations for implementation in at least one device in the data center in connection with the one or more device-related changes by processing the input information and the obtained telemetry data using one or more artificial intelligence techniques; and performing at least one automated action based at least in part on the one or more determined device configurations. | 2021-11-11 |
20210350214 | CONVOLUTIONAL NEURAL NETWORK COMPUTING METHOD AND SYSTEM BASED ON WEIGHT KNEADING - Disclosed embodiments relate to a convolutional neural network computing method and system based on weight kneading, comprising: arranging original weights in a computation sequence and aligning by bit to obtain a weight matrix, removing slack bits in the weight matrix, allowing essential bits in each column of the weight matrix to fill the vacancies according to the computation sequence to obtain an intermediate matrix, removing null rows in the intermediate matrix, obtain a kneading matrix, wherein each row of the kneading matrix serves as a kneading weight; obtaining positional information of the activation corresponding to each bit of the kneading weight; divides the kneading weight by bit into multiple weight segments, processing summation of the weight segments and the corresponding activations according to the positional information, and sending a processing result to an adder tree to obtain an output feature map by means of executing shift-and-add on the processing result. | 2021-11-11 |
20210350215 | TRAINING WITH ADAPTIVE RUNTIME AND PRECISION PROFILING - A mechanism is described for facilitating efficient training of neural networks at computing devices. A method of embodiments, as described herein, includes detecting one or more inputs for training of a neural network, and introducing randomness in floating point (FP) numbers to prevent overtraining of the neural network, where introducing randomness includes replacing less-significant low-order bits of operand and result values with new low-order bits during the training of the neural network. | 2021-11-11 |
20210350216 | ARCHITECTURE ESTIMATION DEVICE, ARCHITECTURE ESTIMATION METHOD, AND COMPUTER READABLE MEDIUM - A reception unit ( | 2021-11-11 |
20210350217 | ANALOG NEURAL MEMORY ARRAY IN ARTIFICIAL NEURAL NETWORK WITH SOURCE LINE PULLDOWN MECHANISM - Numerous embodiments of analog neural memory arrays are disclosed. Certain embodiments contain improved mechanisms for pulling source lines down to ground expeditiously. This is useful, for example, to minimize the voltage drop for a read, program, or erase operation. | 2021-11-11 |
20210350218 | BAYESIAN NEURAL NETWORK WITH RESISTIVE MEMORY HARDWARE ACCELERATOR AND METHOD FOR PROGRAMMING THE SAME - A Bayesian neural network including an input layer, and, an output layer, and, possibly, one or more hidden layer(s). Each neuron of a layer is connected at its input with a plurality of synapses, the synapses of the plurality being implemented as a RRAM array constituted of cells, each column of the array being associated with a synapse and each row of the array being associated with an instance of the set of synaptic coefficients, the cells of a row of the RRAM being programmed during a SET operation with respective programming current intensities, the programming intensity of a cell being derived from the median value of a Gaussian component obtained by GMM decomposition into Gaussian components of the marginal posterior probability of the corresponding synaptic coefficient, once the BNN model has been trained on a training dataset. | 2021-11-11 |
20210350219 | CRESTED BARRIER DEVICE AND SYNAPTIC ELEMENT - A crested barrier memory device may include a first electrode, a first self- rectifying layer, and a combined barrier and active layer. The first self-rectifying layer may be between the first electrode and the active layer. A conduction band offset between the first self-rectifying layer and the combined barrier and active layer may be greater than approximately 1.5 eV. A valence band offset between the first self-rectifying layer and the combined barrier and active layer may be less than approximately −0.5 eV. The device may also include a second electrode. The active layer may be between the first self-rectifying layer and the second electrode. | 2021-11-11 |
20210350220 | HIERARCHICAL HIGHLY HETEROGENEOUS DISTRIBUTED SYSTEM BASED DEEP LEARNING APPLICATION OPTIMIZATION FRAMEWORK - The present invention discloses a hierarchical highly heterogeneous distributed system based deep learning application optimization framework and relates to the field of deep learning in the direction of computational science. The hierarchical highly heterogeneous distributed system based deep learning application optimization framework comprises a running preparation stage and a running stage. The running preparation stage is used for performing deep neural network training. The running stage performs task assignment to all kinds of devices in the distributed system and uses a data encryption module to perform privacy protection to user sensitive data. Due to heterogeneous characteristics of a system task of the present invention, on the premise that the overall performance is guaranteed, the system response time is reduced, the user experience is guaranteed, the data encryption module based on the neural network can perform privacy protection to user sensitive data at a lower computing cost and storage cost, and the user data security is guaranteed. | 2021-11-11 |
20210350221 | Neural Network Inference and Training Using A Universal Coordinate Rotation Digital Computer - A system and method of implementing a neural network with a non-linear activation function is disclosed. A Universal Coordinate Rotation Digital Computer (CORDIC) is used to implement the activation function. Advantageously, the CORDIC is also used during training for back propagation. Using a CORDIC, activation functions such as hyperbolic tangent and sigmoid may be implemented without the use of a multiplier. Further, the derivatives of these functions, which are needed for back propagation, can also be implemented using the CORDIC. | 2021-11-11 |
20210350222 | SYSTEM AND METHOD FOR SELF-SUPERVISED DEPTH AND EGO-MOTION OVERFITTING - Systems and methods to improve machine learning by explicitly over-fitting environmental data obtained by an imaging system, such as a monocular camera are disclosed. The system includes training self-supervised depth and pose networks in monocular visual data collected from a certain area over multiple passes. Pose and depth networks may be trained by extracting data from multiple images of a single environment or trajectory, allowing the system to overfit the image data. | 2021-11-11 |
20210350223 | DIGITAL CONTENT VARIATIONS VIA EXTERNAL REACTION - A method for performing an iteration of an output of a trained GAN may be provided. The method comprises receiving an object as input for the GAN, determining a set of features of the input by the generator adversarial network, generating, by the GAN, at least one modification to one feature of the set of features of the object, generating as output of the GAN the received object as a basis, and the generated modification building a modified object, capturing a feedback signal, and receiving the feedback signal as input by the GAN in a feedback loop for a next iteration. Moreover, the method comprises repeating the determination of a set of features, the generation of at least one modification, the generation of the output and the caption of the feedback signal in the next iteration, wherein as object the modified object is used as the object. | 2021-11-11 |
20210350224 | METHODS AND SYSTEMS FOR EVALUATING A NEW APPLICATION - A computer-implemented method may include: receiving a plurality of documents, wherein each of the plurality of documents includes a description of an application and an objective of the application; extracting the objective of the application from each of the received plurality of documents; clustering the plurality of documents to generate at least a first cluster of documents and a second cluster of documents; determining a first set of attributes associated with each of the documents included in the first cluster of documents and a second set of attributes associated with each of the documents included in the second cluster of documents; and training a classification engine based on the plurality of documents, the extracted objectives, the first set of attributes, and the second set of attributes. | 2021-11-11 |
20210350225 | DETERMINING MULTIVARIATE TIME SERIES DATA DEPENDENCIES - Techniques regarding multivariate time series data analysis are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise a time series analysis component that generates a machine learning model that discovers a dependency between multivariate time series data using an attention mechanism controlled by an uncertainty measure. | 2021-11-11 |
20210350226 | Systems and Methods for Training a Neural Network - Embodiments of the present disclosure include systems and methods for training neural networks. In one embodiment, neural network may receive input data and produce output results in response to the input data and weights of the neural network. An error is determined at an output of the neural network based on the output results. The error is propagated in a reverse direction through the neural network from the output and one or more intermediate outputs to adjust the weights. | 2021-11-11 |
20210350227 | HIGH-RISK PASSAGE AUTOMATION IN A DIGITAL TRANSACTION MANAGEMENT PLATFORM - A document execution engine receives a training set of data including training documents that each include one or more passages associated with a passage type and a level of risk. The document execution engine trains a machine learned model based on the training set. The trained machine learned model, when applied to subsequently identified passages within documents in the document execution environment, can identify a passage with above threshold levels of risk (e.g., a high-risk passage) based on a passage type of the passage. The trained machine learned model can then provide for display the high-risk passage and a related passage of the same passage type from a second document within the document execution environment to the user via a document passage comparison interface. Differences between the passages can be highlighted, enabling a user to quickly compare and contrast the passages. | 2021-11-11 |
20210350228 | PROCESSING METHOD AND APPARATUS OF NEURAL NETWORK MODEL - The disclosure provides a processing method and an apparatus of a neural network model, and relates to a field of computer technologies. The method includes: obtaining and converting input data of the i | 2021-11-11 |
20210350229 | TRAINING TEXT SUMMARIZATION NEURAL NETWORKS WITH AN EXTRACTED SEGMENTS PREDICTION OBJECTIVE - Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a text summarization neural network. One of the methods includes pre-training the text summarization neural network including learning values of a plurality of network parameters through self-supervised learning using unlabeled data comprising unlabeled first texts, the pre-training including: obtaining an unlabeled first text comprising a plurality of segments; selecting one or more of the plurality of segments; processing a masked first text that excludes the one or more selected segments to generate a prediction of the one or more selected segments; and determining, based on a difference between the prediction and the one or more selected segments, an update to the current values of the plurality of network parameters; adapting the pre-trained text summarization neural network for a specific text summarization task using labeled data comprising second texts and respective summaries of the second texts. | 2021-11-11 |
20210350230 | Data dividing method and processor for convolution operation - A data dividing method applied to a computing device that performs a convolution operation based on an input image data and a weight data is provided. The method includes: determining a restriction condition in connection with the performing of the convolution operation by the computing device; determining a set of candidate data blocks for the input image data and a set of candidate data blocks for the weight data according to the restriction condition; generating an evaluation result by evaluating, according to candidate data blocks in the set of candidate data blocks for the input image data and the set of candidate data blocks for the weight data, an amount of data load of the computing device in accessing both an external memory and an internal memory of the computing device; and determining a method of dividing the input image data and the weight data according to the evaluation result. | 2021-11-11 |
20210350231 | PREDICTING A STATE OF A COMPUTER-CONTROLLED ENTITY - A computer-implemented method for enabling control or monitoring of a computer-controlled entity operating in an environment by predicting a future state of the computer-controlled entity and/or its environment using sensor data which is indicative of a current state of the computer-controlled entity and/or its environment. The method includes using a first neural network for approximating a drift component of a stochastic differential equation and a second neural network for approximating a diffusion component of the stochastic differential equation, and discretizing the stochastic differential equation into time steps, and obtaining time-evolving mean and covariance functions based on the discretization and determining a probability distribution of a second state of the computer-controlled entity and/or its environment therefrom. The control of the computer-controlled entity may thus be enhanced and made more efficient and reliable using the uncertainty information available from the determined probability distribution. | 2021-11-11 |
20210350232 | FAULT DETECTION IN CYBER-PHYSICAL SYSTEMS - Methods and systems for training a neural network model include processing a set of normal state training data and a set of fault state training data to generate respective normal state inputs and fault state inputs that each include data features and sensor correlation graph information. A neural network model is trained, using the normal state inputs and the fault state inputs, to generate a fault score that provides a similarity of an input to the fault state training data and an anomaly score that provides a dissimilarity of the input to the normal state training data. | 2021-11-11 |
20210350233 | System and Method for Automated Precision Configuration for Deep Neural Networks - There is provided a system and method of automated precision configuration for deep neural networks. The method includes obtaining an input model and one or more constraints associated with an application and/or target device or process used in the application configured to utilize a deep neural network; learning an optimal low-precision configuration of the architecture using constraints, the training data set, and the validation data set; and deploying the optimal configuration on the target device or process for use in the application. | 2021-11-11 |
20210350234 | TECHNIQUES TO DETECT FUSIBLE OPERATORS WITH MACHINE LEARNING - Various embodiments are generally directed to techniques to detect fusible operators with machine learning, such as by evaluating a set of operators in a graph of a machine learning model to identify fusion candidates comprising subgraphs of the graph with two or more operators to combine, for instance. Some embodiments are particularly directed to utilizing a machine learning classifier to evaluate fusion candidates using a set of features of the fusion candidate. | 2021-11-11 |
20210350235 | SYSTEM AND METHOD FOR HORTICULTURE VIABILITY PREDICTION AND DISPLAY - A system and method is disclosed for plant maintenance, identification and viability scoring. An application is located on consumer devices connected to the server. The application, operating on a smart phone, utilizes onboard GPS, user input, and camera subsystems to customize plant care tips specific to a yard location and plant type. Images may be submitted to the server to identify a plant type through convolutional neural network image recognition. The invention uses another artificial neural network to predict a plant's viability score. The server receives input, such as plant type, soil type, yard location, and amount of sunlight, and the server retrieves local climactic data and plant type optimal values to return the plant's viability score for the selected location. Another aspect of the invention generates and displays an augmented reality display of a plant in the user's yard. | 2021-11-11 |
20210350236 | NEURAL NETWORK ROBUSTNESS VIA BINARY ACTIVATION - A method of increasing neural network robustness. The method comprises defining an artificial neural network comprising a number of bounded ramp activation functions. The network is trained iteratively in a layer-by-layer fashion. Each iteration increases the slope of the activation functions toward a discrete threshold activation and stops when the activation functions converge to the threshold activation and the network exhibits spiking behavior. Alternatively, weight agnostic neural networks are created, wherein nodes in the networks comprise fixed shared weights. A subset of networks is identified that comprise activation functions compatible with neuromorphic hardware and are tested with a specified number of shared weight values. A score is generated for each combination of network and weight value according to performance and mapping to neuromorphic hardware, and the networks are ranked. The networks are then combined according to ranking to create a new network that exhibits spiking behavior. | 2021-11-11 |
20210350237 | System and Method for using Signal Waveform Analysis for Detecting a Change in a Wired Network - An analyzer for monitoring a configuration of a wired network medium that is used for communication between multiple devices. The configuration change includes an additional device tapping to the medium for eavesdropping, or the substituting one of the devices. The analyzer is connected to the medium for receiving, storing, and analyzing waveforms of the physical-layer signals propagated over the medium. The analysis includes comparing the received signals to reference signals, and notifying upon detecting a difference according to pre-set criteria. The analysis may be time or frequency-domain based, and may use a feed-forward Artificial Neural Network (ANN). The wired network may be an automotive or in-vehicle network, PAN, LAN, MAN, or WAN, may use balanced or unbalanced signaling, and may be configured as point-to-point or multi-point topology. The analyzer may be connected at an end of the medium, and may be integrated with one of the devices. | 2021-11-11 |
20210350238 | FAST NEURAL NETWORK IMPLEMENTATIONS BY INCREASING PARALLELISM OF CELL COMPUTATIONS - The amount of time required to train a neural network may be decreased by modifying the neural network to allow for greater parallelization of computations. The computations for cells of the neural network may be modified so that the matrix-vector multiplications of the cell do not depend on a previous cell and thus allowing the matrix-vector computations to be performed outside of the cells. Because the matrix-vector multiplications can be performed outside of the cells, they can be performed in parallel to decrease the computation time required for processing a sequence of training vectors with the neural network. The trained neural network may be applied to a wide variety of applications, such as performing speech recognition, determining a sentiment of text, determining a subject matter of text, answering a question in text, or translating text to another language. | 2021-11-11 |
20210350239 | Non-Uniform Regularization in Artificial Neural Networks for Adaptable Scaling - A system for flexible regularization and adaptable scaling of an artificial neural network is provided. The system includes a memory to store an artificial neural network and training data, a processor and interface to submit signals and training data into the neural network having a sequence of layers, each layer includes a set of neuron nodes, wherein a pair of nodes from neighboring layers are mutually connected with a plural of trainable parameters to pass the signals from the previous layer to next layer, a random number generator to modify the output signal of each neuron nodes for regularization in a stochastic manner following a multi-dimensional distribution across layer depth and node width directions of the neural network, wherein at least one layer has non-identical profile across neuron nodes, a training operator to update the neural network parameters by using the training data such that the output of neural network provides better values in a plural of objective functions; and an adaptive truncator to prune the output of neuron nodes at each layer in a compressed size of the neural network to reduce the computational complexity on the fly in downstream testing phase for any new incoming data. | 2021-11-11 |
20210350240 | SYSTEM AND METHOD FOR COMPRESSING ACTIVATION DATA - A method for adapting a trained neural network is provided. Input data is input to the trained neural network and a plurality of filters are applied to generate a plurality of channels of activation data. Differences between corresponding activation values in the plurality of channels of activation data are calculated and an order of the plurality of channels is determined based on the calculated differences. The neural network is adapted so that it will output channels of activation data in the determined order. The ordering of the channels of activation data is subsequently used to compress activation data values by taking advantage of a correlation between activation data values in adjacent channels. | 2021-11-11 |
20210350241 | APPARATUS AND METHOD FOR SEARCHING FOR A NEURAL NETWORK ARCHITECTURE - An apparatus and method for searching a neural network architecture may be disclosed. The apparatus may include an architecture searcher and an architecture evaluator. The architecture searcher may search for a topology between nodes included in a basic cell of a network, search for an operation to be applied between the nodes after searching for the topology, and determine the basic cell. The architecture evaluator may evaluate performance of the determined basic cell. | 2021-11-11 |
20210350242 | SYSTEMS AND METHODS FOR PRUNING BINARY NEURAL NETWORKS GUIDED BY WEIGHT FLIPPING FREQUENCY - Various embodiments of a system and method for pruning binary neural networks by analyzing weight flipping frequency and pruning the binary neural network based on the weight flipping frequency associated with each channel of the binary neural network are disclosed herein. | 2021-11-11 |
20210350243 | Image Quality Assessment Using Similar Scenes as Reference - A system for image quality assessment of non-aligned images includes a first deep path portion of a convolutional neural network having a set of parameters and a second deep path portion of the convolutional neural network sharing a set of parameters with the first deep path convolutional neural network. Weights are shared between the first and second deep path convolutional neural networks to support extraction of a same set of features in each neural network pathway. Non-aligned reference and distorted images are respectively provided to the first and second deep paths of the convolutional neural network for processing. A concatenation layer is connected to both the first and second deep paths convolutional neural network, and a fully connected layer is connected to the concatenation layer to receive input from both the first and second deep paths of the convolutional neural network, generating an image quality assessment as a linear regressor and outputting an image quality score. | 2021-11-11 |
20210350244 | ATTENTION NEURAL NETWORKS WITH LOCALITY-SENSITIVE HASHING - Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing a machine learning task on a network input to generate a network output. In one aspect, one of the systems includes an attention neural network configured to perform the machine learning task, the attention neural network including one or more LSH attention layers, each LSH attention layer comprising one or more LSH attention sub-layers, each LSH sub-layer configured to: receive a sequence of queries derived from an input sequence to the LSH attention layer, the sequence of queries having a respective query at each of a plurality of input positions; determine one or more respective hash values for each of the respective queries at each of the plurality of input positions; generate a plurality of LSH groupings; and generate an attended input sequence. | 2021-11-11 |