Machine Learning Patents (Class 706/12)
  • Patent number: 11126893
    Abstract: Systems and methods of the present disclosure provide processes for determining how much to adjust machine-learning parameter values in a direction of a gradient for gradient-descent steps in training processes for machine-learning models. Current parameter values of a machine-learning model are vector components that define an initial estimate for a local extremum of a cost function used to measure how well the machine-learning model performs. The initial estimate and the gradient of the cost function for the initial estimate are used to define an auxiliary function. A root estimate is determined for the auxiliary function of the gradient. The parameters are adjusted in the direction of the gradient by an amount specified by the root estimate.
    Type: Grant
    Filed: May 4, 2018
    Date of Patent: September 21, 2021
    Assignee: INTUIT, INC.
    Inventor: William T. Laaser
  • Patent number: 11128564
    Abstract: The system and methods discussed herein provide for filtering out noisy application signatures to improve the precision of first packet application classification. In some implementations, the system receive application signatures from devices along with their network identifiers. Based upon the frequency at which identical application signatures appear as originating from distinct network environments, the system determines the validity of application signatures and avoids storing irrelevant information for routing network traffic.
    Type: Grant
    Filed: November 20, 2018
    Date of Patent: September 21, 2021
    Assignee: Citrix Systems, Inc.
    Inventors: Siddharth G R, Naveen Chowdary Yerramneni, Tarun Kumar Hukmichand
  • Patent number: 11128579
    Abstract: Current chatbot systems cannot understand enough different topics to converse with students who have questions about college admissions, financial aid, courses, and other topics on the path to and through college. Current chatbots also have a hard time understanding misspellings, slang, and context-specific language, e.g., like the language used by students. “Learning” new topics is very time-consuming for current chatbots. And it is difficult for administrators to participate in student conversations carried out in part by current chatbots. To address these technical problems, an inventive chatbot uses a natural language processor (e.g., a neural network) to receive, classify, and respond to queries on thousands of different topics. An inventive chatbot can also request real-time assistance from an administrator when faced with a difficult query and learn on-the-fly from the administrator's response.
    Type: Grant
    Filed: September 29, 2017
    Date of Patent: September 21, 2021
    Assignee: AdmitHub PBC
    Inventors: Andrew Magliozzi, Charles DeTar, Brandon Horst, Peter Corey, Buruk Aregawi
  • Patent number: 11126695
    Abstract: A polymer design device according to an embodiment of the present disclosure receives a requirement for a target physical property of a desired polymer, and acquires structural information of polymers. For each polymer corresponding to the acquired structural information, the polymer design device estimates physical property information of the polymer including a mean value and a standard deviation, based on the structural information of the polymer and a regression model, and calculates a score of the polymer based on the requirement for the target physical property and based on the mean value and the standard deviation. From among the acquired structural information of the polymers, the polymer design device selects at least one polymer as the desired polymer, based on the score of each of the polymers, and outputs information of the selected at least one polymer.
    Type: Grant
    Filed: October 29, 2019
    Date of Patent: September 21, 2021
    Assignee: SHOWA DENKO K.K.
    Inventors: Takuya Minami, Yoshishige Okuno, Katsumi Murofushi, Toshio Fujita
  • Patent number: 11127062
    Abstract: Systems and methods including one or more processing modules and one or more non-transitory storage modules storing computing instructions configured to run on the one or more processing modules and perform acts of training a source classifier with labeled source training data of a first product category from a website of an online retailer, clustering target data for a second product category into a plurality of clusters, inserting into each cluster labeled source training data of the first product category, assigning a domain discriminator score to each cluster, determining whether each cluster comprises an agreement cluster or a disagreement cluster using the domain discriminator score, receiving a product search request for a product of the second category from a user of the web site, and coordinating a display of the product on the web site to promote the product.
    Type: Grant
    Filed: January 23, 2017
    Date of Patent: September 21, 2021
    Assignee: WALMART APOLLP, LLC
    Inventors: Richard Edward Chatwin, Jaymin Daniel Mankowitz, Shie Mannor, Vineet Abhishek
  • Patent number: 11126921
    Abstract: Techniques disclosed herein relate to learning and applying contextual patient similarities. Multiple template similarity functions may be provided. Each template similarity function may compare a respective subset of features of a query entity feature vector with a corresponding subset of features of a candidate entity feature vector. A composite similarity function may be provided as a weighted combination of respective outputs of the template similarity functions. A plurality of labeled entity vectors may be provided as context training data. An approximation function may be applied to approximate a first context label for each respective labeled entity vector. A first context specific composite similarity function may be trained based on the composite similarity function by learning first context weights for the template similarity functions using a first loss function based on output of application of the approximation function to the first context training data.
    Type: Grant
    Filed: April 19, 2018
    Date of Patent: September 21, 2021
    Assignee: KONINKLIJKE PHILIPS N.V.
    Inventors: Bryan Conroy, Minnan Xu, Asif Rahman, Cristhian Mauricio Potes Blandon
  • Patent number: 11126172
    Abstract: Described herein are methods, apparatuses, and systems for reducing equipment repair time. In one embodiment, a computer implemented method includes collecting test substrate data or other metrology data and fault detection data for maintenance recovery of at least one manufacturing tool in a manufacturing facility and determining a relationship between tool parameter settings for the manufacturing tool and the test substrate data. The method further includes utilizing virtual metrology predictive algorithms and at least some collected data to obtain a metrology prediction and applying multivariate run-to-run (R2R) control modeling to obtain a state estimation including a current operating region of the at least one manufacturing tool. Applying multivariate run-to-run (R2R) control modeling to obtain tool parameter adjustments for at least one manufacturing tool to reduce maintenance recovery time and to reduce requalification time.
    Type: Grant
    Filed: August 12, 2019
    Date of Patent: September 21, 2021
    Assignee: Applied Materials, Inc.
    Inventors: Jimmy Iskandar, Jianping Zou, Parris C. M. Hawkins, James Moyne
  • Patent number: 11126503
    Abstract: Techniques are provided for pre-filtering of join execution over multi-column range summaries and other synopses. An exemplary method comprises maintaining a synopsis for a plurality of data tables, wherein a given synopsis summarizes a set of records in a corresponding data table; and, in response to a request for a join operation for a set of the data tables: joining the synopses associated with the set of data tables to generate a joined synopsis; for joined records in the joined synopsis, obtaining corresponding records from the set of data tables as candidate records; and joining the candidate records. Two or more of the set of data tables can be distributed across a plurality of nodes and the synopses can be replicated and/or broadcasted across the plurality of nodes. Incremental updates to broadcasted and/or replicated synopses are optionally provided to at least one node.
    Type: Grant
    Filed: August 10, 2016
    Date of Patent: September 21, 2021
    Assignee: International Business Machines Corporation
    Inventors: Yuan Chi Chang, Timothy Ray Malkemus, Mohammad Sadoghi Hamedani
  • Patent number: 11120102
    Abstract: Systems and methods of determining a global model are provided. In particular, one or more local updates can be received from a plurality of user devices. Each local update can be determined by the respective user device based at least in part on one or more data examples stored on the user device. The one or more data examples stored on the plurality of user devices are distributed on an uneven basis, such that no user device includes a representative sample of the overall distribution of data examples. The local updates can then be aggregated to determine a global model.
    Type: Grant
    Filed: August 27, 2020
    Date of Patent: September 14, 2021
    Assignee: Google LLC
    Inventors: Hugh Brendan McMahan, Jakub Konecny, Eider Brantly Moore, Daniel Ramage, Blaise H. Aguera-Arcas
  • Patent number: 11120344
    Abstract: In various embodiments, a natural language (NL) application implements functionality that enables users to more effectively access various data storage systems based on NL requests. As described, the operations of the NL application are guided by, at least in part, on one or more templates and/or machine-learning models. Advantageously, the templates and/or machine-learning models provide a flexible framework that may be readily tailored to reduce the amount of time and user effort associated with processing NL requests and to increase the overall accuracy of NL application implementations.
    Type: Grant
    Filed: July 29, 2017
    Date of Patent: September 14, 2021
    Assignee: SPLUNK INC.
    Inventors: Dipock Das, Dayanand Pochugari, Neeraj Verma, Nikesh Padakanti, Aungon Nag Radon, Anand Srinivasabagavathar, Adam Oliner
  • Patent number: 11120366
    Abstract: Methods and systems may provide for technology to conduct a machine learning analysis of data access statistics with respect to a plurality of separate datasets and determine a time-dependent access pattern based on the machine learning analysis, wherein the time-dependent access pattern includes an expert access trend, a curation access trend and a knowledgebase access trend. The technology may also generate one or more data management recommendations with response to the plurality of separate datasets based on the time-dependent access pattern.
    Type: Grant
    Filed: July 24, 2018
    Date of Patent: September 14, 2021
    Assignee: International Business Machines Corporation
    Inventors: Patrick Dantressangle, Simon Laws, David George Radley
  • Patent number: 11120053
    Abstract: A computer system includes one or more processors; and one or more non-transitory memories including computer program code, the one or more memories and the computer program code being configured to, with the one or more processors, cause the computer system to perform operations comprising: providing a first set of data records and a second set of data records in which each data record potentially relates to information associated with at least one transitional object; identifying a set of labelings in which at least one label refers to at least one data record; assigning a likelihood score that an identified label corresponds to a data record that is referred to; determining an identity of the at least one transitional object based on the assigned likelihood score; and outputting the determined identity.
    Type: Grant
    Filed: October 23, 2018
    Date of Patent: September 14, 2021
    Assignee: International Business Machines Corporation
    Inventors: Shiau Hong Lim, Laura Wynter
  • Patent number: 11122145
    Abstract: The present approach relates to the use of time series analyses to estimate times or time intervals when a user of IT resources is likely to schedule or request that an operation is run on those services. In certain implementations, the present approach performs forecasting using time series data and supervised machine learning techniques. These techniques may be used to help predict future times when an operation or operations may be requested for execution. Based on these predicted future time, automations (e.g., the automated execution of operations) may be scheduled so as to effectively utilize available resources and efficiently perform the operations.
    Type: Grant
    Filed: October 31, 2019
    Date of Patent: September 14, 2021
    Assignee: ServiceNow, Inc.
    Inventors: Khashayar Goudarzi, Vinayak Raju Kuraku, Sharath Vaddempudi
  • Patent number: 11122069
    Abstract: Devices and methods for detecting a compromised social media account are disclosed. A method includes: receiving, by a computing device, social media content corresponding to a plurality of social media accounts; determining, by the computing device, a plurality of affinity groups, each including two or more social media accounts from the plurality of social media accounts, based upon the received social media content; determining, by the computing device, whether or not a particular social media account of the plurality of social media accounts is compromised using the received social media content and the determined plurality of affinity groups; and in response to determining that the particular social media account is compromised, the computing device providing a notification indicating that the particular social media account is compromised.
    Type: Grant
    Filed: November 26, 2019
    Date of Patent: September 14, 2021
    Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATION
    Inventors: Paul A. R. Frank, Martin G. Keen, Hernan A. Cunico, Adam Smye-Rumsby
  • Patent number: 11122464
    Abstract: A system and method uses a data correlator system to associate large volumes of related data. The data correlator system receives data packets with user content from a mobile network. The data correlator system creates a hash value for data associated with the user content, such as user account information or the location of the user device. The data correlator system uses the hash value to combine the user content with the associated data. The data correlator system provides the combined data to an optimization engine for optimizing the performance of the mobile network.
    Type: Grant
    Filed: August 27, 2019
    Date of Patent: September 14, 2021
    Assignee: AT&T Intellectual Property I, L.P.
    Inventors: Peter Aung Htut Wong, Baofeng Jiang
  • Patent number: 11113612
    Abstract: Computer-implemented systems and methods forecast network resource and/or infrastructure needs for an enterprise computer system that employs network servers to host resources that are requested by network users. Based on the forecasts, the network resources can be scaled or provisioned accordingly. The state of the network servers can be dynamically adjusted to meet the request needs of the users while reducing excess capacity. The forecasting techniques of the present invention are also applicable to cloud computing environments, Based on the forecasts, the cloud server pool can be scaled dynamically, so that the system's scale satisfies the changing requests and avoids wasting resources when the system is under low load.
    Type: Grant
    Filed: December 26, 2016
    Date of Patent: September 7, 2021
    Assignee: MORGAN STANLEY SERVICES GROUP INC.
    Inventors: Reginald W. Martin, Hongbin Zhang, Jian Cao
  • Patent number: 11113624
    Abstract: A distributed machine learning framework implemented with heterogeneous data platforms reduces data copying and exploits memory/computation resources of the different data platforms. A configuration component includes information to set up the system. A persistency component manages storage of data and a model trained by machine learning. A proxy repository includes predefined proxies for communication between heterogeneous data platform nodes and execution of the machine learning procedure. A machine learning execution component comprises three layers. A bottom work node layer within the data platform performs computations of the machine learning procedure. A middle server node layer comprising one server node per data platform, communicates with the work nodes to coordinate jobs on that data platform. An upper layer comprises a central server node communicating with server nodes and coordinating jobs of the different platforms.
    Type: Grant
    Filed: July 12, 2017
    Date of Patent: September 7, 2021
    Assignee: SAP SE
    Inventors: Chengyu Liu, Lian Yang, Xingtian Shi
  • Patent number: 11112994
    Abstract: A system including a memory device with microbumps are disclosed. A non-volatile memory device stores data for a machine learning operation. The non-volatile memory device comprises a set of microbumps. The set of microbumps are to transmit the data for the machine learning operation from the non-volatile memory device to another set of microbumps of a machine learning processing device that performs the machine learning operation.
    Type: Grant
    Filed: December 4, 2019
    Date of Patent: September 7, 2021
    Assignee: Micron Technology, Inc.
    Inventor: Poorna Kale
  • Patent number: 11113398
    Abstract: A mismatch between model-based classifications produced by a first version of a machine learning threat discernment model and a second version of a machine learning threat discernment model for a file is detected. The mismatch is analyzed to determine appropriate handling for the file, and taking an action based on the analyzing. The analyzing includes comparing a human-generated classification status for a file, a first model version status that reflects classification by the first version of the machine learning threat discernment model, and a second model version status that reflects classification by the second version of the machine learning threat discernment model. The analyzing can also include allowing the human-generated classification status to dominate when it is available.
    Type: Grant
    Filed: March 9, 2020
    Date of Patent: September 7, 2021
    Assignee: Cylance Inc.
    Inventors: Kristopher William Harms, Renee Song, Raj Rajamani, Braden Rusell, Yoojin Sohn, Kiefer Ipsen
  • Patent number: 11113603
    Abstract: Some embodiments provide a method for configuring a machine-trained (MT) network that includes input nodes, output nodes, and interior nodes between the input and output nodes. Each node produces an output value and each interior node and output node receives as input values a set of output values of other nodes and applies weights to each received input value. The weights are configurable parameters for training. The method propagates a set of inputs through the MT network to generate a set of outputs. Each input has a corresponding expected output. The method calculates a value of a continuously-differentiable augmented loss function that combines a measurement of a difference between each output and its corresponding expected output and a term that biases training of the weights towards a set of discrete values. The method trains the weights by backpropagating a gradient of the continuously-differentiable augmented loss function at the calculated value.
    Type: Grant
    Filed: November 16, 2017
    Date of Patent: September 7, 2021
    Assignee: PERCEIVE CORPORATION
    Inventors: Steven L. Teig, Eric A. Sather
  • Patent number: 11113337
    Abstract: Embodiments herein provide a method for imputing sensor data, in a sensor data sequence with missing data based on the semantics learning, where semantics is defined by the constraints of the sensor data features. A candidate value for imputation is determined based on sensor data of corresponding instances of time instants of the sensor data sequence using learning based on semantics of features of the sensor data sequence with missing data. The nearest neighbors search has been applied in similar response data sequence using the data values corresponding to the time instant of missing data in sensor data sequence. In case similar response data sequence is not available imputation is performed based on the distribution pattern of missing data.
    Type: Grant
    Filed: September 8, 2017
    Date of Patent: September 7, 2021
    Assignees: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY, TATA CONSULTANCY SERVICES
    Inventors: Soma Bandyopadhyay, Krithivasan Ramamritham
  • Patent number: 11107242
    Abstract: In various examples there is an apparatus for detecting position and orientation of an object. The apparatus comprises a memory storing at least one frame of captured sensor data depicting the object. The apparatus also comprises a trained machine learning system configured to receive the frame of the sensor data and to compute a plurality of two dimensional positions in the frame. Each predicted two dimensional position is a position of sensor data in the frame depicting a keypoint, where a keypoint is a pre-specified 3D position relative to the object. At least one of the keypoints is a floating keypoint depicting a pre-specified position relative to the object, lying inside or outside the object's surface. The apparatus comprises a pose detector which computes the three dimensional position and orientation of the object using the predicted two dimensional positions and outputs the computed three dimensional position and orientation.
    Type: Grant
    Filed: March 22, 2019
    Date of Patent: August 31, 2021
    Assignee: Microsoft Technology Licensing, LLC
    Inventors: Andrew William Fitzgibbon, Erroll William Wood, Jingjing Shen, Thomas Joseph Cashman, Jamie Daniel Joseph Shotton
  • Patent number: 11106758
    Abstract: Methods, systems and computer program products for providing a customized display of social media content are provided. Aspects include receiving a plurality of social media posts that have been published for display by a social media service. Aspects also include receiving a profile associated with a user of the social media service including social media post filtering preferences. The social media post filtering preferences are created based at least in part upon an analysis of social media posts that were previously disliked by the user. Aspects further include identifying a set of acceptable social media posts and a set of unacceptable social media posts. Aspects also include causing the set of acceptable social media posts to be displayed in a news feed window of the social media service in association with an account of the user of the social media service.
    Type: Grant
    Filed: February 7, 2019
    Date of Patent: August 31, 2021
    Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATION
    Inventors: Yuk L. Chan, Michael L. Greenblatt, Heidi Lagares-Greenblatt, Deepti M. Naphade
  • Patent number: 11108710
    Abstract: Disclosed are systems and methods for improving interactions with and between computers in content providing, searching and/or hosting systems supported by or configured with devices, servers and/or platforms. The disclosed systems and methods provide a novel messaging framework that automatically applies a multi-factor analysis technique to incoming and received messages in order to properly identify a message's type and category, which dictates the manner in which the message is displayed within a recipient's inbox. The disclosed framework operates on two levels: i) it determines whether a message is from a human or machine sender (H/M classification), and ii) it determines the messages category (MAGMA categorization).
    Type: Grant
    Filed: January 28, 2020
    Date of Patent: August 31, 2021
    Assignee: VERIZON MEDIA INC.
    Inventors: Neeti Narayan, Hongwei Shang, Changsung Kang, Jean-Marc Langlois
  • Patent number: 11108889
    Abstract: Automatically determining, with reduced (or no) input from the users of a group, a set of activity instances that the group of users has interest in performing. A representation of the set of activity instances can be rendered for consideration by a group, and the set of activity instances can be determined even when only limited criteria are specified. Optionally, in response to affirmative user interface input(s) directed to a rendered representation of the set of activity instances, one or more of the activity instances of the set can be confirmed through limited input(s) of one or more users of the group. Further, the automatic determination of the set of activity instances is optionally performed using one or more trained machine learning models that are trained to optimize a likelihood that the users of the group will find the set satisfactory.
    Type: Grant
    Filed: August 22, 2018
    Date of Patent: August 31, 2021
    Assignee: GOOGLE LLC
    Inventors: Marcos Calvo Lance, Philip Koonce
  • Patent number: 11106789
    Abstract: Anomalous sequences are detected by approximating user sessions with heuristically extracted event sequences, allowing behavior analysis even without user identification or session identifiers. Extraction delimiters may include event count or event timing constraints. Event sequences extracted from logs or other event lists are vectorized and embedded in a vector space. A machine learning model similarity function measures anomalousness of a candidate sequence relative to a specified history, thus computing an anomaly score. Restrictions may be placed on the history to focus on a particular IP address or time frame, without retraining the model. Anomalous sequences may generate alerts, prompt investigations by security personnel, trigger automatic mitigation, trigger automatic acceptance, trigger tool configuration actions, or result in other cybersecurity actions.
    Type: Grant
    Filed: March 5, 2019
    Date of Patent: August 31, 2021
    Assignee: Microsoft Technology Licensing, LLC
    Inventors: Naama Kraus, Roy Levin, Andrey Karpovsky, Tamer Salman
  • Patent number: 11105958
    Abstract: An example method comprises receiving first historical meso-scale numerical weather predictions (NWP) and power flow information for a geographic distribution area, correcting for overfitting of the historical NWP predictions, reducing parameters in the first historical NWP predictions, training first power flow models using the first reduced, corrected historical NWP predictions and the historical power flow information for all or parts of the first geographic distribution area, receiving current NWP predictions for the first geographic distribution area, applying any number of first power flow models to the current NWP predictions to generate any number of power flow predictions, comparing one or more of the any number of power flow predictions to one or more first thresholds to determine significance of reverse power flows, and generating a first report including at least one prediction of the reverse power flow and identifying the first geographic distribution area.
    Type: Grant
    Filed: December 28, 2018
    Date of Patent: August 31, 2021
    Assignee: Utopus Insights, Inc.
    Inventors: Srivats Shukla, Younghun Kim, Aijun Deng
  • Patent number: 11108646
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training and using machine learning models to classify network traffic as IoT traffic or non-IoT traffic and managing the traffic based on the classification. In some implementations, machine learning parameters of a local machine learning model trained by the edge device is received each of at least a subset of a set of edge devices. The machine learning parameters received from an edge device are parameters of the local machine learning model trained by the edge device based on local network traffic processed by the edge device and to classify the network traffic as Internet of Things (IoT) traffic or non-IoT traffic. A global machine learning model is generated, using the machine learning parameters, to classify network traffic processed by edge devices as IoT traffic or non-IoT traffic.
    Type: Grant
    Filed: December 31, 2019
    Date of Patent: August 31, 2021
    Assignee: Hughes Network Systems, LLC
    Inventors: Amit Arora, Satyajit Roy, John Kenyon
  • Patent number: 11105957
    Abstract: A method, system and computer program product are disclosed for integrating plural modalities of information to obtain values for a specified attribute of a given system. In one embodiment, the method comprises acquiring data of a first modality, conveying a first source of data of a first type of the system; configuring simulator with settings of physical sensors; acquiring data of a second modality from the system, conveying a second source of data of a second type of the system. The method further comprises converting the data of the second modality to data of the first type, while configuring a virtual set of sensors to enable acquisition of the converted data of the second modality; and configuring adjoints equipped simulator with settings of the virtual sensors, to mimic collection of data of the first type, while configured to measure data of second type.
    Type: Grant
    Filed: February 20, 2020
    Date of Patent: August 31, 2021
    Assignees: International Business Machines Corporation, Shell Oil Company
    Inventors: Andrew R. Conn, Sippe Douma, Gijs Van Essen, Lior Horesh, Eduardo Antonio Arismendi Jimenez, Ulisses Mello
  • Patent number: 11107027
    Abstract: A method for utilizing an externally augmented propensity model for determining a future financial requirement. The method includes obtaining at least one propensity model that models how data associated with a business entity relates to a future financial requirement of the business entity, and gathering the data associated with the business entity. The data includes a first portion created based on a platform utilized by users associated with the business entity, and financial data of an owner of the business entity. The data matches at least a subset of the at least one propensity model. The business entity is scored by applying the at least one propensity model to the data. Further, based on the score, the future financial requirement of the business entity is classified. Still yet, a message is transmitted to the business entity based on the classification of the future financial requirement of the business entity.
    Type: Grant
    Filed: May 31, 2016
    Date of Patent: August 31, 2021
    Assignee: Intuit Inc.
    Inventors: Eva Diane Chang, Madhu Shalini Iyer, Jeffrey Lewis Kaufman
  • Patent number: 11109105
    Abstract: Systems and methods relate to display of insights, including a system including a remote computing system configured to cause a user display to display a graphical user interface (GUI), wherein the GUI enables a user to select a time range and a locale, the GUI displays a list of a plurality of media content that have been displayed on one or more digital displays associated with the locale, each of the plurality of media content being listed according to impression counts of each of the plurality of media content for the locale selected and the time range selected, and each of the impression counts is a count of views by people of a respective media content of the plurality media content on the one or more digital displays of the locale.
    Type: Grant
    Filed: December 9, 2019
    Date of Patent: August 31, 2021
    Assignee: SHARP NEC DISPLAY SOLUTIONS, LTD.
    Inventors: Richard Ventura, Hiroyuki Kasuga, Steven Keith Platt
  • Patent number: 11100158
    Abstract: Various embodiments provide for selecting a subset of features to use to train a model for search applications. To select a feature, the candidate features are randomly assigned into two groups. Each of the two groups represents a summation of the respective features that were assigned to it. Then a decision tree building scan is performed on the two groups to determine which of the two groups performs better based a selection criteria. Upon determining which of the two groups is better, the candidate features of the winning group are again randomly assigned into two groups. These two groups are again scanned as described above to determine a winning group. This binary splitting and scanning pattern is continuously performed until the winning group contains one remaining feature. That remaining feature is then designated as a selected feature to be used in the search model.
    Type: Grant
    Filed: April 30, 2019
    Date of Patent: August 24, 2021
    Assignee: A9.COM, INC.
    Inventors: Cuize Han, Nikhil Rao, Daria Sorokina, Karthik Subbian
  • Patent number: 11100420
    Abstract: A record extraction request for a data set is received at a machine learning service. A plan to perform one or more chunk-level operations (such as sampling, shuffling, splitting or partitioning for parallel computation) on chunks of the data set is generated. A set of data transfers that results in a particular chunk being stored in a particular server's memory is initiated to implement the first chunk-level operation of the sequence. A second operation such as another filtering operation or a feature processing operation is performed on a result set of the first chunk-level operation.
    Type: Grant
    Filed: August 14, 2014
    Date of Patent: August 24, 2021
    Assignee: Amazon Technologies, Inc.
    Inventors: Leo Parker Dirac, Jin Li, Rakesh Ramakrishnan, Tianming Zheng, Donghui Zhuo
  • Patent number: 11100398
    Abstract: An electronic device may determine whether a machine-learning model is operating within predefined limits. In particular, the electronic device may receive, from another electronic device, instructions for the machine-learning model, a reference input and a predetermined output of the machine-learning model for the reference input. Note that the instructions may include an architecture of the machine-learning model, weights associated with the machine-learning model and/or a set of pre-processing transformations for use when executing the machine-learning model on images. In response, the electronic device may configure the machine-learning model based on the instructions. Then, the electronic device may calculate an output of the machine-learning model for the reference input. Next, the electronic device may determine whether the machine-learning model is operating within predefined limits based on the output and the predetermined output.
    Type: Grant
    Filed: June 30, 2016
    Date of Patent: August 24, 2021
    Assignee: Cogniac, Corp.
    Inventors: William S Kish, Huayan Wang, Sandip C. Patel
  • Patent number: 11102225
    Abstract: One embodiment of the present invention sets forth a technique for predicting fraud by correlating user behavior biometric data with one or more other types of data. The technique includes receiving cursor movement data generated via a client device and analyzing the cursor movement data based on a model to generate a result. The model may be generated based on cursor movement data associated with a first group of one or more users. The technique further includes receiving log data generated via the client device and determining, based on the result and the log data, that a user of the client device is not a member of the first group.
    Type: Grant
    Filed: April 17, 2017
    Date of Patent: August 24, 2021
    Assignee: SPLUNK INC.
    Inventors: Gleb Esman, Oleg Izmerly
  • Patent number: 11100374
    Abstract: A processor-implemented classification method includes: determining a first probability vector including a first probability, for each of a plurality of classes, resulting from a classification of an input with respect to the classes; determining, based on the determined first probability vector, whether one or more of the classes represented in the first probability vector are confusing classes; adjusting, in response to one or more of the classes being the confusing classes, the determined first probability vector based on a first probability of each of the confusing classes and a maximum value of the first probabilities; determining a second probability vector including a second probability, for each of the classes, resulting from another classification of the input with respect to the classes; and performing classification on the input based on a result of a comparison between the determined second probability vector and the adjusted first probability vector.
    Type: Grant
    Filed: November 1, 2019
    Date of Patent: August 24, 2021
    Assignee: Samsung Electronics Co., Ltd.
    Inventors: Young-Seok Kim, Hwidong Na, Seongmin Ok, Min-Joong Lee
  • Patent number: 11100184
    Abstract: Internet searches sometimes provide search results referencing webpages that do not contain all search term elements submitted by a user. The user may then click on such Internet search results where referenced webpages, and/or their descendants, do not contain important search term elements. Also, advertisements are sometimes placed on search results webpages that relate to the user's search term elements, even though some of those search terms are missing in referenced and/or descendant webpages. The present invention is directed to: annotating Internet search results to indicate missing search term elements on referenced and descendant webpages; optionally filtering out search results referencing webpages with missing terms; and showing advertisements related to search term elements.
    Type: Grant
    Filed: December 7, 2020
    Date of Patent: August 24, 2021
    Inventor: Robert Osann, Jr.
  • Patent number: 11100411
    Abstract: A method, system and computer program product are disclosed for predicting influence in a social network. In one embodiment, the method comprises identifying a set of users of the social network, and identifying a subset of the users as influential users based on defined criteria. A multitude of measures are identified as predictors of which ones of the set of users are the influential users. These measures are aggregated, and a composite predictor model is formed based on this aggregation. This composite predictor model is used to predict which ones of the set of users will have a specified influence in the social network in the future. In one embodiment, the specified influence is based on messages sent from the users, and for example, may be based on the number of the messages sent from each user that are re-sent by other users.
    Type: Grant
    Filed: May 25, 2017
    Date of Patent: August 24, 2021
    Assignee: International Business Machines Corporation
    Inventors: Richard D. Lawrence, Estepan Meliksetian, Prem Melville, Claudia Perlich, Karthik Subbian
  • Patent number: 11100116
    Abstract: A computer-implemented method for implementing separated attention on like and dislike items for personalized ranking includes performing an element-wise product on a user embedding and a final like item embedding to generate a first vector. The method further includes performing an element-wise product on the user embedding and a final dislike item embedding to generate a second vector. The method further includes computing a probability that the user prefers the like item to the dislike item based on the first and second vectors, and generating one or more item recommendations including one or more electronic images for the user using the probability.
    Type: Grant
    Filed: October 30, 2018
    Date of Patent: August 24, 2021
    Assignee: International Business Machines Corporation
    Inventors: Shiwan Zhao, Zhi Qiao, Zhi Hu Wang, Li Zhang, Zhong Su
  • Patent number: 11101043
    Abstract: A facility for predicting patient outcomes on the basis of clinical trials is described. The facility obtains information describing one or more completed clinical trials, and extracts features from the obtained clinical trial information. The facility uses the extracted features to train both a time-series data model for predicting clinical outcomes and a non-time-series data model for predicting clinical outcomes. The facility applies these trained models to information describing a subject patient to predict a clinical outcome for the subject patient.
    Type: Grant
    Filed: September 23, 2019
    Date of Patent: August 24, 2021
    Assignee: Zasti Inc.
    Inventors: Ramanathan Krishnan, John Domenech, Rajagopal Jagannathan, Sharath Makki Shankaranarayana
  • Patent number: 11093864
    Abstract: A computing system computes a variable relevance using a trained tree model. (A) A next child node is selected. (B) A number of observations associated with the next child node is computed. (C) A population ratio value is computed. (D) A next leaf node is selected. (E) First observations are identified. (F) A first impurity value is computed for the first observations. (G) Second observations are identified when the first observations are associated with the descending child nodes. (H) A second impurity value is computed for the second observations. (I) A gain contribution is computed. (J) A node gain value is updated. (K) (D) through (J) are repeated. (L) A variable gain value is updated for a variable associated with the split test. (M) (A) through (L) are repeated. (N) A set of relevant variables is selected based on the variable gain value.
    Type: Grant
    Filed: November 10, 2020
    Date of Patent: August 17, 2021
    Assignee: SAS Institute Inc.
    Inventor: Brandon Michael Reese
  • Patent number: 11089985
    Abstract: Behavioral and mental health therapy systems in accordance with several embodiments of the invention include a wearable camera and/or a variety of sensors (accelerometer, microphone, among various other) connected to a computing system including a display, audio output, holographic output, and/or vibrotactile output to automatically recognize social cues from images captured by at least one camera and provide this information to the wearer via one or more outputs such as (but not limited to) displaying an image, displaying a holographic overlay, generating an audible signal, and/or generating a vibration.
    Type: Grant
    Filed: October 9, 2020
    Date of Patent: August 17, 2021
    Assignees: The Board of Trustees of the Leland Stanford Junior University
    Inventors: Catalin Voss, Nicholas Joseph Haber, Dennis Paul Wall, Aaron Scott Kline, Terry Allen Winograd
  • Patent number: 11093751
    Abstract: A system and methods are disclosed for using a trained machine learning model to identify constituent images within composite images. A method may include providing pixel data of a first image as input to the trained machine learning model, obtaining one or more outputs from the trained machine learning model, and extracting, from the one or more outputs, a level of confidence that (i) the first image is a composite image that includes a constituent image, and (ii) at least a portion of the constituent image is in a particular spatial area of the first image.
    Type: Grant
    Filed: March 9, 2020
    Date of Patent: August 17, 2021
    Assignee: GOOGLE LLC
    Inventors: Filip Pavetic, King Hong Thomas Leung, Dmitrii Tochilkin
  • Patent number: 11093852
    Abstract: A system of classifying devices and/or app instances a new or returning divides attributes generated from observations received from an uncharacterized device/software app into base-fingerprint attributes and predictor attributes, where the two kinds of attributes have different longevities. Predictor attribute tuples from attribute tuples having the same base fingerprint as the base fingerprint corresponding to the uncharacterized device/app, and the predictor attribute tuple corresponding to the uncharacterized device/app are analyzed using a machine learned predictor function to obtain a final fingerprint. Machine learning techniques such as logistic regression, support vector machine, and artificial neural network can provide a predictor function that can decrease the conflict rate of the final fingerprint and, hence, the utility thereof, without significantly affecting the accuracy of classification.
    Type: Grant
    Filed: October 19, 2016
    Date of Patent: August 17, 2021
    Assignee: ACCERTIFY, INC.
    Inventors: Glenn S. Benson, Kasun Maduranga Samarasinghe
  • Patent number: 11093566
    Abstract: A method and system for improving a router based search query is provided. The method includes identifying a Web page retrieved during a Web query received from a network router device and analyzing historical Web pages retrieved during historical search queries associated with a search engine router and the network router device. The Web page is ranked with respect to the historical Web pages and the Web page is tagged with a first domain comprising a topic associated with the Web page. An address for the network router device is tagged with the first domain and additional domains comprising additional topics associated with additional Web pages accessed via the network router device. In response, the Web page is re-ranked with respect to the historical Web pages.
    Type: Grant
    Filed: May 20, 2019
    Date of Patent: August 17, 2021
    Assignee: International Business Machines Corporation
    Inventors: Ashrith Shetty, Rohit S. Shetty
  • Patent number: 11087234
    Abstract: The present teaching relates to distributed deep machine learning on a cluster. In one example, a request is received for estimating one or more parameters associated with a machine learning model on a cluster including a plurality of nodes. A set of data is obtained to be used for estimating the one or more parameters. The set of data is divided into a plurality of sub-sets of data, each of which corresponds to one of the plurality of nodes. Each sub-set of data is allocated to a corresponding node for estimating values of the one or more parameters based on the sub-set of data. Estimated values of the one or more parameters obtained based on a corresponding sub-set of data allocated to the node, are received from each of the plurality of nodes. The one or more parameters of the machine learning model are estimated based on the estimated values of the one or more parameters generated by at least some of the plurality of nodes.
    Type: Grant
    Filed: January 29, 2016
    Date of Patent: August 10, 2021
    Assignee: Verizon Media Inc.
    Inventors: Andrew Feng, Jun Shi, Mridul Jain, Peter Cnudde
  • Patent number: 11086891
    Abstract: A system for tracking and representing data science data runs includes a hub including a first computing device communicatively coupled with a data store. A runner including a second computing device having a cache is communicatively coupled with the hub through a telecommunications network. An end user computing device includes a display and is communicatively coupled with the runner and the hub. User interfaces displayed on the display include: a unique identifier identifying a data science data run performed by the runner; a list of input files used by the runner to perform the run; a list of output files output by the runner as a result of the run; and a diagram diagramming a process flow including a visual representation of the input files, a visual representation of the run, and a visual representation of the output files.
    Type: Grant
    Filed: January 8, 2020
    Date of Patent: August 10, 2021
    Assignee: Subtree Inc.
    Inventors: Luke Marsden, Alaric Blagrave Snell-Pym, Karolis Rusenas, Charlotte Rachael Godley
  • Patent number: 11087742
    Abstract: Systems and methods are described herein for generating adaptive feedback in response to a user request. Input indicative of a user request may be received and utilized to identify an item in an electronic catalog. A title for the item may be retrieved and provided, as input, to a machine-learning model. The machine-learning model may be trained to identify one or more segments of an input title. A shortened title may be generated from these identified segments and provided as output at the user device (e.g., via audible output provided at a speaker of the user device). In some embodiments, the length and content of the shortened title may vary based at least in part on the contextual intent of the user's request.
    Type: Grant
    Filed: June 12, 2019
    Date of Patent: August 10, 2021
    Assignee: Amazon Technologies, Inc.
    Inventors: Ran Levy, Yehuda Finkelstein, Iftah Gamzu, Haim Litvak
  • Patent number: 11086860
    Abstract: An information retrieval system and method are presented. A template is retrieved from a template repository. The template repository stores a plurality of templates. Each of the plurality of templates includes a concept and a relationship from a knowledge model. The knowledge model defines a plurality of entities and interrelationships between one or more of the plurality of entities. The plurality of entities include concepts and instances. The template is transmitted to a client computer, and a statement is received from the client computer. The statement includes an instantiation of the template. A knowledge base is queried using the statement to generate a result listing identifying an item in the knowledge base. The knowledge base identifies a plurality of items. Each of the plurality of items is associated with at least one annotation identifying at one of the entities in the knowledge model.
    Type: Grant
    Filed: March 14, 2019
    Date of Patent: August 10, 2021
    Assignee: CAPRICORN HOLDINGS PTE, LTD
    Inventors: Sinuhé Arroyo, Carlos Ruiz Moreno
  • Patent number: 11087184
    Abstract: A computer-implemented method and system are provided for training a model for New Class Categorization (NCC) of a test image. The method includes decoupling, by a hardware processor, a feature extraction part from a classifier part of a deep classification model by reparametrizing learnable weight variables of the classifier part as a combination of learnable variables of the feature extraction part and of a classification weight generator of the classifier part. The method further includes training, by the hardware processor, the deep classification model to obtain a trained deep classification model by (i) learning the feature extraction part as a multiclass classification task, and (ii) episodically training the classifier part by learning a classification weight generator which outputs classification weights given a training image.
    Type: Grant
    Filed: September 24, 2019
    Date of Patent: August 10, 2021
    Inventors: Renqiang Min, Kai Li, Bing Bai, Hans Peter Graf