Abstract: An image processing apparatus, includes a memory; and a processor coupled to the memory and configured to: generate a trained machine learning model by learning a machine learning model using a first set of image data, output an inference result by inputting a second set of image data to the trained machine learning model, and process a region of interest at a time of inference with respect to image data for which an inference result is correct in the second set of image data.

Abstract: A convolutional neural network performs: a first masking process of masking a pixel region not to be inspected, by computing pixel values of corresponding pixels of an inspection image and of a mask image; an intermediate process for extracting a feature image from the inspection image that has been subjected to the first masking process; and a second masking process of masking the pixel region not to be inspected, by computing the pixel values of corresponding pixels of the inspection image that has been subjected to the intermediate process and of the mask image that has been subjected to a process identical to the intermediate process.

Abstract: A computer system and method for extending parallelized asynchronous reinforcement learning for training a neural network is described in various embodiments, through coordinated operation of plurality of hardware processors or threads such that each functions as a worker agent that is configured to simultaneously interact with a target computing environment for local gradient computation based on a loss determination and to update global network parameters based at least on local gradient computation to train the neural network through modifications of weighted interconnections between interconnected computing units as gradient computation is conducted across a plurality of iterations of a target computing environment, the loss determination including at least a policy loss term (actor), a value loss term (critic), and an auxiliary control loss. Variations are described further where the neural network is adapted to include terminal state prediction and action guidance.

Abstract: Techniques are provided for extracting anomaly related rules from organizational data. One method comprises obtaining anomaly analysis data integrated from multiple data sources of an organization, wherein the multiple data sources comprise at least one set of labeled anomaly data related to anomalous transactions; extracting features from the integrated anomaly analysis data that correlate with an indication of an anomaly; training multiple machine learning models using the extracted features, where the machine learning models are trained using different combinations of the extracted features; evaluating a performance of the trained machine learning models; and extracting rules from the trained machine learning models based on the performance, wherein the extracted rules are used to classify transactions as anomalous. The trained machine learning models comprise a decision tree comprising paths to an anomaly classification. The extracted rules are optionally in a human-readable format.

Abstract: An apparatus includes first and second compute-in-memory (CIM) arrays. The first CIM array is configured to store weights corresponding to a filter tensor, to receive a first set of activations corresponding to a first receptive field of an input, and to process the first set of activations with the weights to generate a corresponding first tensor of output values. The second CIM array is configured to store a first copy of the weights corresponding to the filter tensor and to receive a second set of activations corresponding to a second receptive field of the input. The second CIM array is also configured to process the second set of activations with the first copy of the weights to generate a corresponding second tensor of output values. The first and second compute-in-memory arrays are configured to process the first and second receptive fields in parallel.

Abstract: Methods, systems, devices and computer software/program code products enable efficiently finding stereo correspondence between a feature or set of features in a first image or signal, and a search domain in a second image or signal.

Abstract: The invention relates to a system for generating synthetic digital data, comprising: a receiver configured to receive at least one measured signal, in particular an RF signal or a sensor signal, a converter configured to convert the at least one measured signal to a digital dataset representing signal characteristics of the at least one measured signal, at least one trainable neural network encoder, wherein, during a training routine, the neural network encoder is configured to receive the digital dataset and to generate a compressed representation of the digital dataset, a processing unit configured to analyze the compressed representation and to detect a correlation between the digital dataset and the compressed representation, wherein the processing unit is configured to generate decoder input data based on the detected correlation, and a trained neural network decoder which is configured to receive the decoder input data and to generate synthetic digital data representing signal characteristics of the at

Abstract: The present disclosure is directed to systems and methods for classifying biomedical images. A feature classifier may generate a plurality of tiles from a biomedical image. Each tile may correspond to a portion of the biomedical image. The feature classifier may select a subset of tiles from the plurality of tiles by applying an inference model. The subset of tiles may have highest scores. Each score may indicate a likelihood that the corresponding tile includes a feature indicative of the presence of the condition. The feature classifier may determine a classification result for the biomedical image by applying an aggregation model. The classification result may indicate whether the biomedical includes the presence or lack of the condition.

Abstract: Provided are various mechanisms and processes for automatic computer vision-based defect detection using a neural network. A system is configured for receiving historical datasets that include training images corresponding to one or more known defects. Each training image is converted into a corresponding matrix representation for training the neural network to adjust weighted parameters based on the known defects. Once sufficiently trained, a test image of an object that is not part of the historical dataset is obtained. Portions of the test image are extracted as input patches for input into the neural network as respective matrix representations. A probability score indicating the likelihood that the input patch includes a defect is automatically generated for each input patch using the weighted parameters. An overall defect score for the test image is then generated based on the probability scores to indicate the condition of the object.

Abstract: An image processing method is provided that automatically calculates Body Surface Area (BSA) score using machine learning techniques. A Felzenszwalb image segmentation algorithm is used to define proposed regions in each of a plurality of training set images. The training set images are oversegmented, and then each of the proposed regions in each of the plurality of oversegmented training set images are manually classified as being a lesion or a non-lesion. A Convolutional Neural Network (CNN) is then trained using the manually classified proposed regions in each of the plurality of training set images. The trained CNN is then used on test images to calculate BSA scores.

Abstract: Embodiments of the present invention provide an improvement to convention machine model training techniques by providing an innovative system, method and computer program product for the generation of synthetic data using an iterative process that incorporates multiple machine learning models and neural network approaches. A collaborative system for receiving data and continuously analyzing the data to determine emerging patterns is provided. Common characteristics of data from the identified emerging patterns are broadened in scope and used to generate a synthetic data set using a generative neural network approach. The resulting synthetic data set is narrowed based on analysis of the synthetic data as compared to the detected emerging patterns, and can then be used to further train one or more machine learning models for further pattern detection.

Abstract: Systems and methods for processing video content are disclosed. According to at least one embodiment, a method of processing video content includes: receiving the video content; identifying a portion of the video content based on one or more of a plurality of factors, the plurality of factors including colour, raster percentage, brightness, and temporal factors; and providing an indicator that facilitates visual recognition of the identified portion of the video content.

Abstract: A deposit detection device according to an embodiment includes a detection module and an identification module. The detection module detects a small region as a candidate region for a deposit region corresponding to a deposit adhering to an imaging device, based on brightness information for each of small regions into which a predetermined region in an image captured by the imaging device is divided. The identification module identifies the candidate region as the deposit region when undulation change in brightness distribution of pixels included in the candidate region detected by the detection module is within a predetermined range.

Abstract: A first image of an electronic device is captured and processed in order to identify the electronic device. After identifying the electronic device, information is downloaded to a mobile device including specific wiring instructions for connecting the electronic device, which are superimposed onto an image of the electronic device to guide a user of the mobile device in wiring the electronic device. A second image of an at least partially wired electronic device may be captured and processed to ascertain whether there are any wiring errors. When wiring errors are found by the remote server, the mobile device receives a message from the remote server indicating the wiring errors that were found.

Abstract: Automated systems and methods for determining the variability between derived expression scores for a series of biomarkers between different identified cell clusters in a whole slide image are presented. The variability between derived expression scores may be a derived inter-marker heterogeneity metric.

Abstract: The presently-disclosed technology enables real-time inspection of a multitude of subcomponents of a component in parallel. For example, the component may be a semiconductor package, and the subcomponents may include through-silicon vias. One embodiment relates to a method for inspecting multiple subcomponents of a component for defects, the method comprising, for each subcomponent undergoing defect detection: extracting a subcomponent image from image data of the component; computing a transformed feature vector from the subcomponent image; computing pairwise distances from the transformed feature vector to each transformed feature vector in a training set; determining a proximity metric using said pairwise distances; and comparing the proximity metric against a proximity threshold to detect a defect in the subcomponent. Another embodiment relates to a product manufactured using a disclosed method of inspecting multiple subcomponents of a component for defects.

Abstract: A method of generating training spectra for training of a neural network includes generating a plurality of theoretically generated initial spectra from an optical model, sending the plurality of theoretically generated initial spectra to a feedforward neural network to generate a plurality of modified theoretically generated spectra, sending an output of the feedforward neural network and empirically collected spectra to a discriminatory convolutional neural network, determining that the discriminatory convolutional neural network does not discriminate between the modified theoretically generated spectra and empirically collected spectra, and thereafter, generating a plurality of training spectra from the feedforward neural network.

Abstract: Implementations are described herein for edge-based real time crop yield predictions made using sampled subsets of robotically-acquired vision data. In various implementations, one or more robots may be deployed amongst a plurality of plants in an area such as a field. Using one or more vision sensors of the one or more robots, a superset of high resolution images may be acquired that depict the plurality of plants. A subset of multiple high resolution images may then be sampled from the superset of high resolution images. Data indicative of the subset of high resolution images may be applied as input across a machine learning model, with or without additional data, to generate output indicative of a real time crop yield prediction.

Abstract: Large scale instance recognition is provided that can take advantage of channel-wise pooling. A received query image is processed to extract a set of features that can be used to generate a set of region proposals. The proposed regions of image data are processed using a trained classifier to classify the regions as object or non-object regions. Extracted features for the object regions are processed using feature correlation against extracted features for a set of object images, each representing a classified object. Matching tensors generated from the comparison are processed using a spatial verification network to determine match scores for the various object images with respect to a specific object region. The match scores are used to determine which objects, or types of objects, are represented in the query image. Information or content associated with the matching objects can be provided as part of a response.

Abstract: According to various embodiments of the disclosure, an electronic device may include an image sensor and a processor. The processor may be configured to detect a movement of an object, using an image generated by the image sensor, to identify a size value of the object, to correct a size value of the object based on a location of the object within the image, and to perform an operation corresponding to a movement of the object, based on the corrected size value.