Google Inc.

Oracle America, Inc.

Sun Microsystems, Inc.

Sacha Christophe Arnoud

Liron Yatziv

Qian Yu

Patrick G Gates

Vinay Damodar Shet

Olaf Manczak

Martin Christian Stumpe

Jeremy Matthew Werner

Sherif Botros

Steven Waterhouse

Dar-Shyang Lee

Oded Fuhrmann

Christian Szegedy

Dumitru Erhan

Kong Man Cheung

Marco Zennaro

Nicholas Aaron Solter

Julian Ibarz

Edison Tan

David John Abraham

Stephane Brossier

Shamim P Mohamed

Eltefaat Shokri

Frederic Barrat

Michael Crawford

Joseph Andrew Mcclain

Colin Mcmillen

Luis Von Ahn

Bryan Arthur Pendleton

Angelique Moscicki

Mark R Russell

Aspects of the disclosure provide for a method for updating business information. A business at a business location and a reference image for the business at the business location may be selected for an update. From a number of more recent images, a comparison image may be selected based on a likelihood that the comparison image captures the business location. Text and visual features reference image may be compared with text and visual features of the comparison image to determine a text similarity score and a features similarity score. A confidence level indicating whether the business in the reference image is in the comparison image may then be determined using the text similarity score and the feature similarity score. According to the confidence level, the business information of the business may be updated.

Business change detection from street level imagery

Aspects of the present disclosure relate to a method includes training a deep neural network using training images and data identifying one or more business storefront locations in the training images. The deep neural network outputs tight bounding boxes on each image. At the deep neural network, a first image may be received. The first image may be evaluated using the deep neural network. Bounding boxes may then be generated identifying business storefront locations in the first image.

Business discovery from imagery

Systems and methods for updating geographic data based on a transaction are provided. In some aspects, one or more transaction records associated with a business are accessed from a memory. Each transaction record identifies a transaction time, geographic location data, and transaction information. A geocoded record of the business is selected to update, based on the geographic location data of the one or more transaction records. The selected geocoded record is updated based on at least one of the transaction time or the transaction information identified in the transaction records.

Updating geographic data based on a transaction

A plurality of cells forming at least a portion of a hive of a data storage system may be capacity balanced by fragmenting a portion of at least one non-empty tile of one of the plurality of cells and moving the fragmented portion to another one of the plurality of cells. A plurality of cells forming at least a portion of a hive of a fixed content storage system may be capacity balanced by identifying at least one of the plurality of cells from which objects are to be moved, and for each of the at least one of the plurality of cells identified, determining a number of objects to be moved to another one of the plurality of cells, identifying one or more tiles that collectively have approximately the number of objects to be moved, and moving the one or more tiles to the another one of the plurality of cells.

Method and system for capacity-balancing cells of a storage system

A node availability management system includes a president node which includes a message queue to store a message, a decision engine configured to generate an operation based on the message, where said operation includes a plurality of atomic actions arranged in a dependency graph, an executor configured to execute the operation by executing the plurality of atomic instructions in the dependency graph, and a slave node operatively connected to the president node, where the executor offloads the execution of at least one of the plurality of atomic actions to the slave node.

Effective high availability cluster management and effective state propagation for failure recovery in high availability clusters

A method for distributively storing a data object on a data storage system includes determining a layout map ID and associating the layout map ID with the data object. The layout map ID is used in determining a layout map, and the layout map is used in determining a layout given a current configuration of the data storage system. The data object is fragmented into data fragments and parity fragments, and the data fragments and parity fragments are distributively stored on data storage system in accordance with the layout.

Method for distributed storage of data

A method includes determining a garbage collection list for a data storage structure of a node in a data storage system, the garbage collection list including one or more layout map identifiers (IDs) for garbage collection. A data fragment stored on the data storage structure is located at a first location and a layout map ID associated with the data fragment is determined. A determination is made whether the layout map ID associated with the data fragment matches a layout map ID for garbage collection in the garbage collection list. If the layout map ID associated with the data fragment matches a layout map ID for garbage collection in the garbage collection list, a determination is made whether the data fragment is present at a second location on the data storage system. If the data fragment is present at a second location on the data storage system, the data fragment at the first location is determined to be a garbage fragment and deleted from the data storage system.

Method of garbage collection on a data storage system

A method for recovery of data objects on a data storage system includes determining a list of layout map IDs that place data objects on a node of a data storage system. For each layout map ID, a list of object IDs associated with each of the data objects assigned the layout map ID is determined. For each object ID, a list of data object fragments associated with the object ID is determined. A determination is made whether each of the data object fragments is present on the node. If a data object fragment is not present on the node, the data object fragment is reconstructed on the data storage system. The method can be automatically implemented by each of one or more operational nodes of a distributed data storage system at configured time intervals, and/or upon the operational loss of a data storage structure of a node, or a node.

Method for recovery of data

Growing community of inventors

San Francisco, CA, United States of America

Sacha Christophe Arnoud