Patents by Inventor Stephen Purcell
Stephen Purcell has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20240095259Abstract: Systems and methods are provided for storing a first data object comprising a first set of immutable components, the first data object being associated with a corresponding second data object stored by a remote replication system. A difference is determined between the first set of immutable components of the first data object and a second set of immutable components of the corresponding second data object. A subset of immutable components is identified from the first set of immutable components based on the difference. The subset of immutable components from the first set of immutable components is provided to the remote replication system over a communication network.Type: ApplicationFiled: November 28, 2023Publication date: March 21, 2024Inventors: Stephen Freiberg, Alexander Landau, Andrew Greene, Brian Dorne, Bryan Offutt, Ernest Zeidman, Ilya Nepomnyaschchiy, John Garrod, Katherine Brainard, Kolin Purcell, Michael Levin, Simon Swanson, Spencer Stecko
-
Publication number: 20230351227Abstract: A modular machine learning-as-a-service (MLAAS) system uses machine learning to respond to tasks without requiring machine learning modeling or design knowledge by its users. The MLAAS system receives an inference request including a model identifier and a target defining features for use in processing the inference request. The features correspond to a task for evaluation using a machine learning model associated with the model identifier. An inference outcome is generated by processing the inference request using the target as input to the model. Feedback indicating an accuracy of the inference outcome with respect to the task is later received and used to generate a training data set, which the MLAAS can use to further train model used to generate the inference outcome. As a result, the training of a machine learning model by the MLAAS system is limited to using data resulting from an inference performed using that model.Type: ApplicationFiled: July 2, 2023Publication date: November 2, 2023Inventors: George BANIS, Adam Starikiewicz, Kevin M. Walsh, Stephen Purcell, Hector Urdiales, Andrea Bergonzo
-
Patent number: 11727287Abstract: A modular machine learning-as-a-service (MLAAS) system uses machine learning to respond to tasks without requiring machine learning modeling or design knowledge by its users. The MLAAS system receives an inference request including a model identifier and a target defining features for use in processing the inference request. The features correspond to a task for evaluation using a machine learning model associated with the model identifier. An inference outcome is generated by processing the inference request using the target as input to the model. Feedback indicating an accuracy of the inference outcome with respect to the task is later received and used to generate a training data set, which the MLAAS can use to further train model used to generate the inference outcome. As a result, the training of a machine learning model by the MLAAS system is limited to using data resulting from an inference performed using that model.Type: GrantFiled: August 8, 2022Date of Patent: August 15, 2023Inventors: George Banis, Adam Starikiewicz, Kevin M. Walsh, Stephen Purcell, Hector Urdiales, Andrea Bergonzo
-
Publication number: 20220383199Abstract: A modular machine learning-as-a-service (MLAAS) system uses machine learning to respond to tasks without requiring machine learning modeling or design knowledge by its users. The MLAAS system receives an inference request including a model identifier and a target defining features for use in processing the inference request. The features correspond to a task for evaluation using a machine learning model associated with the model identifier. An inference outcome is generated by processing the inference request using the target as input to the model. Feedback indicating an accuracy of the inference outcome with respect to the task is later received and used to generate a training data set, which the MLAAS can use to further train model used to generate the inference outcome. As a result, the training of a machine learning model by the MLAAS system is limited to using data resulting from an inference performed using that model.Type: ApplicationFiled: August 8, 2022Publication date: December 1, 2022Inventors: George BANIS, Adam STARIKIEWICZ, Kevin M. WALSH, Stephen PURCELL, Hector URDIALES, Andrea BERGONZO
-
Patent number: 11449775Abstract: A modular machine learning-as-a-service (MLAAS) system uses machine learning to respond to tasks without requiring machine learning modeling or design knowledge by its users. The MLAAS system receives an inference request including a model identifier and a target defining features for use in processing the inference request. The features correspond to a task for evaluation using a machine learning model associated with the model identifier. An inference outcome is generated by processing the inference request using the target as input to the model. Feedback indicating an accuracy of the inference outcome with respect to the task is later received and used to generate a training data set, which the MLAAS can use to further train model used to generate the inference outcome. As a result, the training of a machine learning model by the MLAAS system is limited to using data resulting from an inference performed using that model.Type: GrantFiled: December 17, 2019Date of Patent: September 20, 2022Assignee: HubSpot, Inc.Inventors: George Banis, Adam Starikiewicz, Kevin M. Walsh, Stephen Purcell, Hector Urdiales, Andrea Bergonzo
-
Publication number: 20200394832Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: ApplicationFiled: August 27, 2020Publication date: December 17, 2020Inventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 10789758Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: GrantFiled: December 19, 2018Date of Patent: September 29, 2020Assignee: Imagination Technologies LimitedInventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Publication number: 20200210867Abstract: A modular machine learning-as-a-service (MLAAS) system uses machine learning to respond to tasks without requiring machine learning modeling or design knowledge by its users. The MLAAS system receives an inference request including a model identifier and a target defining features for use in processing the inference request. The features correspond to a task for evaluation using a machine learning model associated with the model identifier. An inference outcome is generated by processing the inference request using the target as input to the model. Feedback indicating an accuracy of the inference outcome with respect to the task is later received and used to generate a training data set, which the MLAAS can use to further train model used to generate the inference outcome. As a result, the training of a machine learning model by the MLAAS system is limited to using data resulting from an inference performed using that model.Type: ApplicationFiled: December 17, 2019Publication date: July 2, 2020Inventors: George Banis, Adam Starikiewicz, Kevin M. Walsh, Stephen Purcell, Hector Urdiales, Andrea Bergonzo
-
Publication number: 20190122419Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: ApplicationFiled: December 19, 2018Publication date: April 25, 2019Inventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 10192350Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: GrantFiled: April 25, 2017Date of Patent: January 29, 2019Assignee: Imagination Technologies LimitedInventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Publication number: 20170228920Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: ApplicationFiled: April 25, 2017Publication date: August 10, 2017Inventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 9633468Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: GrantFiled: December 2, 2014Date of Patent: April 25, 2017Assignee: Imagination Technologies LimitedInventors: Luke T Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Publication number: 20150302630Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: ApplicationFiled: December 2, 2014Publication date: October 22, 2015Inventors: Luke T. Peterson, James A. McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 8988433Abstract: Aspects include systems, methods, and media for implementing methods relating to increasing consistency of results during intersection testing. In an example, vertexes define edges of primitives composing a scene (e.g., triangles defining a mesh for a surface of an object in a 3-D scene). An edge can be shared between two primitives. Intersection testing algorithms can use tests involving edges to determine whether or not the ray intersects a primitive defined by those edges. In one approach, a precedence among the vertexes defining a particular edge is enforced for such intersection testing. The precedence causes an intersection tester to always test a given edge in the same orientation, regardless of which primitive defined (at least in part) by that edge is being intersection tested.Type: GrantFiled: April 28, 2011Date of Patent: March 24, 2015Assignee: Imagination Technologies, LimitedInventors: Stephen Purcell, Christopher Philip Alan Tann, Jason Rupert Redgrave, Cüneyt Özdaş
-
Patent number: 8927942Abstract: A ion source comprises: a chamber, an injection to inject matter into the chamber, wherein said matter comprises at least a first species, a tip with an apex located in the chamber, wherein the apex has a surface made of a metallic second species, a generator to generate ions of said species, and a regulation system adapted to set operative conditions of the chamber to alternatively generate ions from the gaseous first species, and ions from the non-gaseous metallic second species.Type: GrantFiled: June 5, 2012Date of Patent: January 6, 2015Assignees: Centre National de la Recherche Scientifique—CNRS, Universite Claude Bernard LyonInventors: Jacques Gierak, Nicolas Allemandou, Klaus Hasselbach, Jean-Paul Leggeri, Guillaume Donnier-Valentin, Stephen Purcell, Pauline Gavand
-
Patent number: 8902230Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: GrantFiled: August 5, 2013Date of Patent: December 2, 2014Assignee: Imagination Technologies, LimitedInventors: Luke Tilman Peterson, James Alexander McCombe, Ryan R. Salsbury, Stephen Purcell
-
Publication number: 20140175301Abstract: A ion source comprises: a chamber (45), an injection to inject matter into the chamber, wherein said matter comprises at least a first species, a tip with an apex located in the chamber, wherein the apex has a surface made of a metallic second species, a generator to generate ions of said species, and a regulation system adapted to set operative conditions of the chamber to alternatively generate ions from the gaseous first species, and ions from the non-gaseous metallic second species.Type: ApplicationFiled: June 5, 2012Publication date: June 26, 2014Applicants: UNIVERSITE LYON 1 CLAUDE BERNARD, CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE -CNRSInventors: Jacques Gierak, Nicolas Allemandou, Klaus Hasselbach, Jean-Paul Leggeri, Guillaume Donnier-Valentin, Stephen Purcell, Pauline Gavand
-
Publication number: 20140071123Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Responsibility for executing these operations can be distributed among different sets of computation units. The sets of computation units each can execute a set of instructions on a parallelized set of input data elements and produce results. These results can be that the data elements can be categorized into different subsets, where each subset requires different processing as a next step. The data elements of these different subsets can be coalesced so that they are contiguous in a results set. The results set can be used to schedule additional computation, and if there are empty locations of a scheduling vector (after accounting for the members of a given subset), then those empty locations can be filled with other data elements that require the same further processing as that subset.Type: ApplicationFiled: August 5, 2013Publication date: March 13, 2014Applicant: Caustic Graphics, Inc.Inventors: Luke Tilman Peterson, James Alexander McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 8502820Abstract: Ray tracing, and more generally, graphics operations taking place in a 3-D scene, involve a plurality of constituent graphics operations. Scheduling of graphics operations for concurrent execution on a computer may increase throughput. In aspects herein, constituent graphics operations are scheduled in groups, having members selected according to disclosed aspects. Processing for specific graphics operations in a group can be deferred if all the operations in the group cannot be further tested concurrently. Graphics operations that have been deferred are recombined into two or more different groups and ultimately complete processing, through a required number of iterations of such process. In one application, the performance of the graphics operations perform a search in which respective 1:1 matches between different types of geometric shapes involved in the 3-D scene are identified.Type: GrantFiled: June 16, 2012Date of Patent: August 6, 2013Assignee: Caustic Graphics, Inc.Inventors: Luke Tilman Peterson, James Alexander McCombe, Ryan R. Salsbury, Stephen Purcell
-
Patent number: 8421801Abstract: Systems, methods, and computer readable media embodying such methods provide for allowing specification of per-ray clipping information that defines a sub-portion of a 3-D scene in which the ray should be traced. The clipping information can be specified as a clip distance from a ray origin, as an end value of a parametric ray definition, or alternatively the clipping information can be built into a definition of the ray to be traced. The clipping information can be used to check whether portions of an acceleration structure need to be traversed, as well as whether primitives should be tested for intersection. Other aspects include specifying a default object that can be returned as intersected when no primitive was intersected within the sub-portion defined for testing. Further aspects include allowing provision of flags interpretable by an intersection testing resource that control what the intersection testing resource does, and/or what information it reports after conclusion of testing of a ray.Type: GrantFiled: September 8, 2009Date of Patent: April 16, 2013Assignee: Caustic Graphics, Inc.Inventors: Ryan R. Salsbury, James Alexander McCombe, Stephen Purcell, Luke Tilman Peterson