Embodiments of the invention utilize a feature-extraction approach and/or a matching approach in combination with a nonparametric approach to estimate the proportion of documents in each of multiple labeled categories with high accuracy. The feature-extraction approach automatically generates continuously valued text features optimized for estimating the category proportions, and the matching approach constructs a matched set that closely resembles a data set that is unobserved based on an observed set, thereby improving the degree to which the distributions of the observed and unobserved sets resemble each other.

A method includes obtaining multiple local activation times (LATs) at different respective measurement locations on an anatomical surface of a heart. The method further includes computing respective directions of electrical propagation at one or more sampling locations on the anatomical surface, by, for each sampling location, selecting a respective subset of the measurement locations for the sampling location, constructing a set of vectors, each of at least some of the vectors including, for a different respective measurement location in the subset, three position values derived from respective position coordinates of the measurement location and an LAT value derived from the LAT at the measurement location, and computing the direction of electrical propagation at the sampling location based on a Principal Component Analysis (PCA) of a 4×4 covariance matrix for the set of vectors. The method further includes indicating the directions of electrical propagation on a display.

Methods and systems for building and maintaining model(s) of a physical process are disclosed. One method includes receiving training data associated with a plurality of different data sources, and performing a clustering process to form one or more clusters. For each of the one or more clusters, the method includes building a data model based on the training data associated with the data sources in the cluster, automatically performing a data cleansing process on operational data based on the data model, and automatically updating the data model based on updated training data that is received as operational data. For data sources excluded from the clusters, automatic building, data cleansing, and updating of models can also be applied.

A system in a vehicle includes a lidar system to transmit incident light and receive reflections from one or more objects as a point cloud of points. The system also includes processing circuitry to identify feature points among the points of the point cloud, the feature points being horizontal feature points reflected from a horizontal surface or vertical feature points reflected from a vertical surface. The processing circuitry processes the point cloud by obtaining a normal vector corresponding to each of the points of the point cloud. The normal vector includes a first component associated with a first dimension, a second component associated with a second dimension, and a third component associated with a third dimension.

Obfuscating a human or other subject in digital media preserves privacy. A user of a smartphone, for example, may enable a flag for obscuring her face in digital photos or movies. When any device captures digital media, the user's smartphone transmits the flag for receipt. The device capturing the digital media is thus informed of the user's desire to obscure her face or even entire image. The device capturing the digital media may thus perform an obscuration in response to the flag.

A system and method of re-identifying players in a broadcast video feed are provided herein. A computing system retrieves a broadcast video feed for a sporting event. The broadcast video feed includes a plurality of video frames. The computing system generates a plurality of tracks based on the plurality of video frames. Each track includes a plurality of image patches associated with at least one player. Each image patch of the plurality of image patches is a subset of the corresponding frame of the plurality of video frames. For each track, the computing system generates a gallery of image patches. A jersey number of each player is visible in each image patch of the gallery. The computing system matches, via a convolutional autoencoder, tracks across galleries. The computing system measures, via a neural network, a similarity score for each matched track and associates two tracks based on the measured similarity.

Systems, methods and aspects, and embodiments thereof relate to unsupervised or semi-supervised features learning using a quantum processor. To achieve unsupervised or semi-supervised features learning, the quantum processor is programmed to achieve Hierarchal Deep Learning (referred to as HDL) over one or more data sets. Systems and methods search for, parse, and detect maximally repeating patterns in one or more data sets or across data or data sets. Embodiments and aspects regard using sparse coding to detect maximally repeating patterns in or across data. Examples of sparse coding include L0 and L1 sparse coding. Some implementations may involve appending, incorporating or attaching labels to dictionary elements, or constituent elements of one or more dictionaries. There may be a logical association between label and the element labeled such that the process of unsupervised or semi-supervised feature learning spans both the elements and the incorporated, attached or appended label.

The invention provides a method for identifying a characteristic of one or more ultrasound images, wherein each image is of a subject imaged by an ultrasound probe using an ultrasound imaging process. The method includes obtaining a manipulation signal indicative of a manipulation of the ultrasound probe during the imaging process. A portion of the manipulation signal, which indicates the manipulation of the probe during a time period, is obtained. The obtained portion is associated with one or more ultrasound images. A neural network system is then used to classify a characteristic of the one or more ultrasound images based on both the obtained portion of the manipulation signal and the one or more images themselves. Such classification comprises applying one or more convolution kernels on the obtained portion of the manipulation signal to generate a convolution output representative of the obtained portion of the manipulation signal, and classifying the convolution output to indicate the characteristic of the one or more ultrasound images associated with the obtained portion.

A system for marking cardiac time intervals from heart valve signals includes a non-invasive sensor unit for capturing electrical signals and composite vibration objects, a memory containing computer instructions, and one or more processors coupled to the memory. The one or more processors causes the one or more processors to perform operations including separating a plurality of individual heart vibration events into heart valve signals from the composite vibration objects, and marking cardiac time interval from the heart valve signals by detecting individual heartbeats using at least one or more of a PCA algorithm or deep learning.

The present technology analyzes the content of images to create complex representations of the images and then reduces the complexity of these representations into a size that is both suitable for comparison but also contains critical image descriptive aspects. These reduced complexity representations can then be used to efficiently search for similar images. Moreover, the reduced complexity representations are formatted such that they can take advantage of existing text search engines, which are well suited to efficiently searching through a large number of unique results.