公开(公告)号：US11359925B2

公开(公告)日：2022-06-14

申请号：US16573901

申请日：2019-09-17

Applicant: Salesforce.com, Inc.

Inventor： John Brian Stewart ,  Dinesh Premalal Weerapurage ,  Timothy Alan Carnes ,  Christopher Sebastian Groer ,  Lahiru Sandakith Pileththuwasan Gallege ,  Chenlu Lou

IPC: G01C21/34 ,  G01C21/32

Abstract: A method involves accelerating the electronic determination of high quality solutions to routing problems by leveraging simplified travel time approximations. The use of first approximation travel time estimates allows for more rapid beginning of a process for electronically determining one or more high quality solutions to a routing problem as compared to having to wait for computation of second travel time estimates for every ordered pair of locations involved in the routing problem, thus allowing for more rapid determination of one or more high quality solutions.

公开(公告)号：US11356508B1

公开(公告)日：2022-06-07

申请号：US17382201

申请日：2021-07-21

Applicant: salesforce.com, inc.

Inventor： Jeaneth Aguilar Vergara ,  Arifullah Khan Pattan

IPC: G06F15/16 ,  H04L67/1095 ,  H04L67/06

Abstract: Computing systems, for example, multi-tenant systems deploy software artifacts in data centers created in a cloud platform using a cloud platform infrastructure language that is cloud platform independent. The system generates pipelines for deploying software artifacts in data center entities configured in a cloud platform. The system allows partial execution of pipelines such that the pipeline can be executed again to complete execution of the remaining stages. The system maintains state of the pipeline execution and checks the state to determine whether a stage should be executed during subsequent executions. The system allows a failed stage to be retried multiple times based on a retry strategy. A retry strategy may depend on the data center entity in a hierarchy of data venter entities of a data center.

公开(公告)号：US20220171696A1

公开(公告)日：2022-06-02

申请号：US17108875

申请日：2020-12-01

Applicant: salesforce.com, inc.

Inventor： Heng Zhang ,  Utsavi Benani ,  Zhidong Ke

IPC: G06F11/36 ,  G06F8/60

Abstract: Mechanisms and techniques for providing an isolated runtime environment are disclosed. Scoped properties are generated utilizing a namespace identifier. The namespace identifier is an identifier used to define a scope of at least an isolated application runtime context. Scoped objects are generated corresponding to one or more dependent services utilizing the one or more scoped properties and the namespace identifier. A target application is deployed to a container corresponding to the isolated application runtime context. The target application interacts with at least some of the dependent services utilizing the namespace identifier. At least one integration test is run on the target application in the container. The integration tests refer to the isolated application runtime context utilizing the namespace identifier.

公开(公告)号：US11349908B2

公开(公告)日：2022-05-31

申请号：US16687376

申请日：2019-11-18

Applicant: salesforce.com, inc.

Inventor： Aleksey Pershin

IPC: G06F7/00 ,  H04L67/10 ,  H04L67/51 ,  G06F16/22 ,  G06F16/2458

Abstract: A system uses templates to generate data structures, for example, pipelines for deploying services in cloud based system. A template contains templating expressions used as placeholders for actual values. The generated data structure may be modified, for example, using an application. The system generates a reverse substitution map that associates actual values used in a data structure instance with a corresponding templating expression used as placeholders for the actual values. The system generates a modified version of the template for a modified version of the data structure by replacing actual values with the corresponding templating expressions from the reverse substitution map. The system can use the modified template to generate other instances of data structure that include the modifications made to the data structure.

公开(公告)号：US20220164547A1

公开(公告)日：2022-05-26

申请号：US17150988

申请日：2021-01-15

Applicant: salesforce.com, inc.

Inventor： Samson Min Rong Tan ,  Shafiq Rayhan Joty

IPC: G06F40/45 ,  G06N3/08 ,  G06F40/289 ,  G06F40/30

Abstract: Embodiments described herein provide adversarial attacks targeting the cross-lingual generalization ability of massive multilingual representations, demonstrating their effectiveness on multilingual models for natural language inference and question answering. An efficient adversarial training scheme can thus be implemented with the adversarial attacks, which takes the same number of steps as standard supervised training and show that it encourages language-invariance in representations, thereby improving both clean and robust accuracy.

公开(公告)号：US20220161425A1

公开(公告)日：2022-05-26

申请号：US17538741

申请日：2021-11-30

Applicant: Salesforce.com, inc.

Inventor： Reid Carlberg ,  Heather Dykstra ,  Philippe Ozil ,  Alan Dawkuan Hwang ,  Cody Henshaw

IPC: B25J9/16 ,  G06Q50/28 ,  G06N5/02

Abstract: Architectures and techniques to control heterogeneous teams of robots. Input is received from a remote device with a control platform. The control platform provides a control mechanism for a team of heterogenous robots having at least two different control structures. Control signals are generated for at least two different types of robots communicatively coupled with the control platform. A first type of robot is independent of a second type of robot and the first and second type of robot have different control mechanisms. The first type of robot receives a request for an item to be delivered. At least one of the first type of robot operates to identify the item and to move the item to an intermediate location and to generate a request to at least one of the second type of robot. At least one of the second type of robot operates to move the item from the intermediate location to a new location.

公开(公告)号：US11341130B2

公开(公告)日：2022-05-24

申请号：US16852395

申请日：2020-04-17

Applicant: salesforce.com, inc.

Inventor： Rohitashva Mathur

IPC: G06F16/2452 ,  G06F16/2455 ,  G06F16/22 ,  G06F16/242

Abstract: A query to retrieve database values for a first subset of table columns in a database table is determined. Runtime column selection variables are set up to identify the first subset of table columns for data retrieval, and are dynamically bound to a prepared query statement with single-column query statements joined by interstitial union operators. Each single-column query statement in the prepared query statement specifies a single-column query directed to a corresponding table column with a corresponding predicate to be bound at runtime dynamically with a corresponding portion of the one or more runtime column selection variables. The predicate is evaluated to be true or false in dependence on whether the corresponding table column is in the first subset of table columns. The prepared query statement is executed to retrieve database values for the first subset of table columns in the database table.

公开(公告)号：US20220156527A1

公开(公告)日：2022-05-19

申请号：US17209011

申请日：2021-03-22

Applicant: salesforce.com, inc.

Inventor： Ramprasaath Ramasamy Selvaraju ,  Nikhil Naik

IPC: G06K9/62 ,  G06T7/194 ,  G06K9/32 ,  G06K9/46

Abstract: Embodiments described herein embodiments described herein provide Contrastive Attention-Supervised Tuning (CAST), a training method to fix the visual grounding ability of contrastive SSL methods based on a data augmentation strategy using unsupervised saliency maps. In addition to the contrastive loss that encourages the model to pick the crop that comes from the corresponding image, CAST provides an explicit grounding supervision through a Grad-CAM based attention loss that enforces models to look at the specified object of interest that is common across different crops when making this decision. A new geometric transform is introduced for randomly cropping different views from an input image based on certain constraints derived from a saliency map.

公开(公告)号：US11334766B2

公开(公告)日：2022-05-17

申请号：US16778339

申请日：2020-01-31

Applicant: salesforce.com, inc.

Inventor： Junnan Li ,  Chu Hong Hoi

IPC: G06K9/62 ,  G06N3/04 ,  G06N3/08

Abstract: Systems and methods are provided for training object detectors of a neural network model with a mixture of label noise and bounding box noise. According to some embodiments, a learning framework is provided which jointly optimizes object labels, bounding box coordinates, and model parameters by performing alternating noise correction and model training. In some embodiments, to disentangle label noise and bounding box noise, a two-step noise correction method is employed. In some examples, the first step performs class-agnostic bounding box correction by minimizing classifier discrepancy and maximizing region objectness. In some examples, the second step uses dual detection heads for label correction and class-specific bounding box refinement.

公开(公告)号：US20220150199A1

公开(公告)日：2022-05-12

申请号：US17583551

申请日：2022-01-25

Applicant: salesforce.com, inc.

Inventor： James Michael Ciancio-Bunch ,  Scott McCorkle ,  Richard W. Jamison

IPC: H04L51/10 ,  H04L51/222 ,  H04L51/234

Abstract: A system for introducing live content into an electronic mail message comprising at least one dynamic live content area sent through an e-mail service provider system comprises memory and a click manager. The memory includes live content comprising image data and expiration data. The click manager receives an indication of the opening of the message by a recipient, wherein the click manager accesses the memory to retrieve live content and send the retrieved live content for rendering in the dynamic live content area of the e-mail message opened by the one of the plurality of recipients.