Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an object embedding system. In one aspect, a method comprises providing selected images as input to the object embedding system and generating corresponding embeddings, wherein the object embedding system comprises a thumbnailing neural network and an embedding neural network. The method further comprises backpropagating gradients based on a loss function to reduce the distance between embeddings for same instances of objects, and to increase the distance between embeddings for different instances of objects.

Methods, systems, and computer program products for enabling distance-based algorithms on data encrypted using a 2DNF homomorphic encryption scheme with inefficient decryption are provided herein. A computer-implemented method includes generating multiple versions of a data point, wherein each of the multiple versions of the data point comprises a distinct value corresponding to a distinct Euclidean space; encrypting each of the multiple versions of the data point; storing the multiple encrypted versions of the data point across multiple databases; and executing one or more distance-based algorithms on the multiple encrypted versions of the data point by using a finite decryption table across the multiple databases, wherein the finite decryption table stores a set of plaintext-ciphertext mappings between (i) multiple plaintext values and (ii) multiple encrypted ciphertext values corresponding to the multiple plaintext values.

Embodiments described herein provide an apparatus comprising a processor to project voxels from a point cloud data set into an n-DoF space, and define successively less granular supervoxels at successively higher layer of abstraction in a view of the point cloud data set, and a memory communicatively coupled to the processor. Other embodiments may be described and claimed.

Systems and methods for analyzing a customer deployment in a converged or hyper-converged infrastructure are disclosed. A machine learning model is trained based upon historical usage data of other customer deployments. A k-means clustering is performed to generate a prediction as to whether a deployment is configured for optimal failover. Recommendations to improve failover performance can also be generated.

Systems and methods for determining a location based on image data are provided. A method can include receiving, by a computing system, a query image depicting a surrounding environment of a vehicle. The query image can be input into a machine-learned image embedding model and a machine-learned feature extraction model to obtain a query embedding and a query feature representation, respectively. The method can include identifying a subset of candidate embeddings that have embeddings similar to the query embedding. The method can include obtaining a respective feature representation for each image associated with the subset of candidate embeddings. The method can include determining a set of relative displacements between each image associated with the subset of candidate embeddings and the query image and determining a localized state of a vehicle based at least in part on the set of relative displacements.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an object embedding system. In one aspect, a method comprises providing selected images as input to the object embedding system and generating corresponding embeddings, wherein the object embedding system comprises a thumbnailing neural network and an embedding neural network. The method further comprises backpropagating gradients based on a loss function to reduce the distance between embeddings for same instances of objects, and to increase the distance between embeddings for different instances of objects.

The invention proposes a method of training a convolutional neural network in which, at each convolution layer, weights for one seed convolutional filter per layer are updated during each training iteration. All other convolutional filters are polynomial transformations of the seed filter, or, alternatively, all response maps are polynomial transformations of the response map generated by the seed filter.

A method for analyzing an anatomical structure of a patient may include the steps of receiving volumetric scan data representative of one or more features of an anatomical structure; mapping the features to a node tree diagram; and displaying the node tree diagram. The features can comprise branching points, pathways connecting the branching points, and location data of the branching points and pathways. The node tree diagram may comprise a plurality of nodes and branches representing the branching points and pathways in the anatomical structure, respectively. The plurality of nodes may comprise a root node representing a root branching point as well as additional nodes representing additional branching points. Additionally, the node tree diagram may comprise a first set of one or more regions, wherein each region encompasses a respective portion of the node tree diagram and is representative of a defined portion of the anatomical structure.

Methods and systems for determining a network embedding include training a network embedding model using training data that includes topology information for networks and attribute information relating to vertices of the networks. An embedded representation is generated using the trained network embedding model to represent an input network, with associated attribute information, in a network topology space. A machine learning task is performed using the embedded representation as input to a machine learning model.

Cameras capture time-stamped images of predefined areas. The images are processed to make decisions as to when a person depicted in the images takes possession of or is disposed of an item depicted in the images. Possessed items are added to a shopping cart maintained for the person and dispossessed items are removed from the shopping cart.