Patents by Inventor Luis Henrique Ceze
Luis Henrique Ceze 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: 20240112089Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: ApplicationFiled: December 13, 2023Publication date: April 4, 2024Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11886963Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: GrantFiled: February 23, 2021Date of Patent: January 30, 2024Assignee: OctoML, Inc.Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11816545Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: GrantFiled: November 9, 2021Date of Patent: November 14, 2023Assignee: OCTOML, INC.Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Publication number: 20220172110Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: ApplicationFiled: February 23, 2021Publication date: June 2, 2022Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Publication number: 20220172119Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: ApplicationFiled: November 9, 2021Publication date: June 2, 2022Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11348036Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: GrantFiled: February 23, 2021Date of Patent: May 31, 2022Assignee: OctoML, Inc.Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11315042Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: GrantFiled: February 23, 2021Date of Patent: April 26, 2022Assignee: OctoML, Inc.Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11216752Abstract: A facility for optimizing machine learning models is described. The facility obtains a description of a machine learning model and a hardware target for the machine learning model. The facility obtains optimization result data from a repository of optimization result data. The facility optimizes the machine learning model for the hardware target based on the optimization result data.Type: GrantFiled: February 23, 2021Date of Patent: January 4, 2022Assignee: OctoML, Inc.Inventors: Matthew Welsh, Jason Knight, Jared Roesch, Thierry Moreau, Adelbert Chang, Tianqi Chen, Luis Henrique Ceze, An Wang, Michal Piszczek, Andrew McHarg, Fletcher Haynes
-
Patent number: 11164661Abstract: In some embodiments, systems and methods for storing and/or retrieving digital information in a nucleic acid library are provided. In some embodiments, an integrated system comprising a nucleic acid synthesis device, a nucleic acid sequencing device, a computing device, and a nucleic acid library is provided. In some embodiments, a write request that associates a value with a key is received by the system, the system synthesizes nucleic acid molecules associated with the request, and stores the nucleic acid molecules in the nucleic acid library. In some embodiments, a read request for a key is received by the system, and the system sequences nucleic acid molecules from the nucleic acid library that are associated with the key.Type: GrantFiled: April 8, 2016Date of Patent: November 2, 2021Assignee: UNIVERSITY OF WASHINGTONInventors: Luis Henrique Ceze, Georg Seelig
-
Publication number: 20180068060Abstract: In some embodiments, systems and methods for storing and/or retrieving digital information in a nucleic acid library are provided. In some embodiments, an integrated system comprising a nucleic acid synthesis device, a nucleic acid sequencing device, a computing device, and a nucleic acid library is provided. In some embodiments, a write request that associates a value with a key is received by the system, the system synthesizes nucleic acid molecules associated with the request, and stores the nucleic acid molecules in the nucleic acid library. In some embodiments, a read request for a key is received by the system, and the system sequences nucleic acid molecules from the nucleic acid library that are associated with the key.Type: ApplicationFiled: April 8, 2016Publication date: March 8, 2018Applicant: University of WashingtonInventors: Luis Henrique Ceze, Georg Seelig
-
Patent number: 9786386Abstract: A memory chip for dynamic approximate storage includes an array of memory cells associated with at least two regions. The chip further includes at least one threshold register for storing values for thresholds for memory cells corresponding to each of the at least two regions; and control logic to programmatically adjust the values for the thresholds for the memory cells. A method of controlling a storage device for dynamic approximate storage includes modifying at least one value stored in a threshold register and associated with at least one cell in a region of a memory comprising at least two regions to apply a biasing for the at least one cell, wherein the biasing adjusts ranges for values in a cell.Type: GrantFiled: February 27, 2015Date of Patent: October 10, 2017Assignee: Microsoft Technology Licensing, LLCInventors: Karin Strauss, Luis Henrique Ceze, Henrique S. Malvar, Qing Guo
-
Patent number: 9690656Abstract: A method of encoding data on single level or variable multi-level cell storage includes receiving a block of encoded data from an approximation-aware application and at least an importance attribute associated with the block of encoded data; and assigning the block of encoded data to a memory address or a particular region of a memory having at least three precision levels, based at least according to the importance attribute. The importance attribute indicates a relative sensitivity of bits of the block to errors in an output quality from decoding the encoded data. An approximation-aware application can be an image encoding application having a modified entropy encoding step that enables identification and splitting of bits into groupings according to sensitivity to errors.Type: GrantFiled: February 27, 2015Date of Patent: June 27, 2017Assignee: Microsoft Technology Licensing, LLCInventors: Karin Strauss, Luis Henrique Ceze, Henrique S. Malvar, Qing Guo
-
Publication number: 20160254063Abstract: A memory chip for dynamic approximate storage includes an array of memory cells associated with at least two regions. The chip further includes at least one threshold register for storing values for thresholds for memory cells corresponding to each of the at least two regions; and control logic to programmatically adjust the values for the thresholds for the memory cells. A method of controlling a storage device for dynamic approximate storage includes modifying at least one value stored in a threshold register and associated with at least one cell in a region of a memory comprising at least two regions to apply a biasing for the at least one cell, wherein the biasing adjusts ranges for values in a cell.Type: ApplicationFiled: February 27, 2015Publication date: September 1, 2016Inventors: KARIN STRAUSS, LUIS HENRIQUE CEZE, HENRIQUE S. MALVAR, QING GUO
-
Publication number: 20160253238Abstract: A method of encoding data on single level or variable multi-level cell storage includes receiving a block of encoded data from an approximation-aware application and at least an importance attribute associated with the block of encoded data; and assigning the block of encoded data to a memory address or a particular region of a memory having at least three precision levels, based at least according to the importance attribute. The importance attribute indicates a relative sensitivity of bits of the block to errors in an output quality from decoding the encoded data. An approximation-aware application can be an image encoding application having a modified entropy encoding step that enables identification and splitting of bits into groupings according to sensitivity to errors.Type: ApplicationFiled: February 27, 2015Publication date: September 1, 2016Inventors: KARIN STRAUSS, LUIS HENRIQUE CEZE, HENRIQUE S. MALVAR, QING GUO
-
Patent number: 9412466Abstract: The present technology relaxes the precision (or full data-correctness-guarantees) requirements in memory operations, such as writing or reading, of MLC memories so that an application may write and read a digital data value as an approximate value. Types of MLCs include Flash MLC and MLC Phase Change Memory (PCM) as well as other resistive technologies. Many software applications may not need the accuracy or precision typically used to store and read data values. For example, an application may render an image on a relatively low resolution display and may not need an accurate data value for each pixel. By relaxing the precision or correctness requirements is a memory operation, MLC memories may have increased performance, lifetime, density, and/or energy efficiency.Type: GrantFiled: September 25, 2014Date of Patent: August 9, 2016Assignee: MICROSOFT TECHNOLOGY LICENSING, LLCInventors: Karin Strauss, Douglas C. Burger, Luis Henrique Ceze, Adrian Sampson
-
Patent number: 9146746Abstract: Devices and methods for providing deterministic execution of multithreaded applications are provided. In some embodiments, each thread is provided access to an isolated memory region, such as a private cache. In some embodiments, more than one private cache are synchronized via a modified MOESI coherence protocol. The modified coherence protocol may be configured to refrain from synchronizing the isolated memory regions until the end of an execution quantum. The execution quantum may end when all threads experience a quantum end event such as reaching a threshold instruction count, overflowing the isolated memory region, and/or attempting to access a lock released by a different thread in the same quantum.Type: GrantFiled: March 1, 2012Date of Patent: September 29, 2015Assignee: University of Washington through its Center of CommercializationInventors: Luis Henrique Ceze, Thomas Bergan, Joseph Devietti, Daniel Joseph Grossman, Jacob Eric Nelson
-
Patent number: 9021313Abstract: A system and method are provided for enhancing approximate computing by a computer system. In one example, an interface is provided comprising a variable-identifier module and a bit-priority module. The variable-identifier module is configured to identify one or more variables of data that are to be processed by the computer system with approximate precision. Approximate precision is a precision level at which a hardware device does not guarantee full data-correctness for the one or more variables. The bit-priority module is configured to assign bit-priorities to the one or more variables. The bit-priorities include relative levels of importance among bits of each of the one or more variables. The relative levels of importance include at least high-priority bits and low-priority bits.Type: GrantFiled: November 28, 2012Date of Patent: April 28, 2015Assignee: Microsoft Technology Licensing, LLCInventors: Karin Strauss, Adrian Sampson, Luis Henrique Ceze
-
Publication number: 20150009736Abstract: The present technology relaxes the precision (or full data-correctness-guarantees) requirements in memory operations, such as writing or reading, of MLC memories so that an application may write and read a digital data value as an approximate value. Types of MLCs include Flash MLC and MLC Phase Change Memory (PCM) as well as other resistive technologies. Many software applications may not need the accuracy or precision typically used to store and read data values. For example, an application may render an image on a relatively low resolution display and may not need an accurate data value for each pixel. By relaxing the precision or correctness requirements is a memory operation, MLC memories may have increased performance, lifetime, density, and/or energy efficiency.Type: ApplicationFiled: September 25, 2014Publication date: January 8, 2015Inventors: Karin Strauss, Douglas C. Burger, Luis Henrique Ceze, Adrian Sampson
-
Patent number: 8861270Abstract: The present technology relaxes the precision (or full data-correctness-guarantees) requirements in memory operations, such as writing or reading, of MLC memories so that an application may write and read a digital data value as an approximate value. Types of MLCs include Flash MLC and MLC Phase Change Memory (PCM) as well as other resistive technologies. Many software applications may not need the accuracy or precision typically used to store and read data values. For example, an application may render an image on a relatively low resolution display and may not need an accurate data value for each pixel. By relaxing the precision or correctness requirements is a memory operation, MLC memories may have increased performance, lifetime, density, and/or energy efficiency.Type: GrantFiled: March 11, 2013Date of Patent: October 14, 2014Assignee: Microsoft CorporationInventors: Karin Strauss, Adrian Sampson, Luis Henrique Ceze, Douglas C. Burger
-
Publication number: 20140258593Abstract: The present technology relaxes the precision (or full data-correctness-guarantees) requirements in memory operations, such as writing or reading, of MLC memories so that an application may write and read a digital data value as an approximate value. Types of MLCs include Flash MLC and MLC Phase Change Memory (PCM) as well as other resistive technologies. Many software applications may not need the accuracy or precision typically used to store and read data values. For example, an application may render an image on a relatively low resolution display and may not need an accurate data value for each pixel. By relaxing the precision or correctness requirements is a memory operation, MLC memories may have increased performance, lifetime, density, and/or energy efficiency.Type: ApplicationFiled: March 11, 2013Publication date: September 11, 2014Applicant: Microsoft CorporationInventors: Karin Strauss, Adrian Sampson, Luis Henrique Ceze, Douglas C. Burger