Patents by Inventor David Ha
David Ha 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).
-
Patent number: 11978141Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating images. In one aspect, a method includes: receiving an input text prompt including a sequence of text tokens in a natural language; processing the input text prompt using a text encoder neural network to generate a set of contextual embeddings of the input text prompt; and processing the contextual embeddings through a sequence of generative neural networks to generate a final output image that depicts a scene that is described by the input text prompt.Type: GrantFiled: May 19, 2023Date of Patent: May 7, 2024Assignee: Google LLCInventors: Chitwan Saharia, William Chan, Mohammad Norouzi, Saurabh Saxena, Yi Li, Jay Ha Whang, David James Fleet, Jonathan Ho
-
Patent number: 11932695Abstract: Provided herein are modified antibodies, compounds used to make them, and intermediates in their synthesis; compositions; formulations and methods, including methods of treating diseases, disorders or conditions, for example, cancer, in humans.Type: GrantFiled: June 17, 2020Date of Patent: March 19, 2024Assignee: Sensei Biotherapeutics, Inc.Inventors: David Y. Jackson, Edward Ha
-
Patent number: 11900235Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing inputs using recurrent neural networks. One of the systems includes a main recurrent neural network comprising one or more recurrent neural network layers and a respective hyper recurrent neural network corresponding to each of the one or more recurrent neural network layers, wherein each hyper recurrent neural network is configured to, at each of a plurality of time steps: process the layer input at the time step to the corresponding recurrent neural network layer, the current layer hidden state of the corresponding recurrent neural network layer, and a current hypernetwork hidden state of the hyper recurrent neural network to generate an updated hypernetwork hidden state.Type: GrantFiled: September 9, 2021Date of Patent: February 13, 2024Assignee: Google LLCInventors: Andrew M. Dai, Quoc V. Le, David Ha
-
Publication number: 20240046106Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using multi-task neural networks. One of the methods includes receiving a first network input and data identifying a first machine learning task to be performed on the first network input; selecting a path through the plurality of layers in a super neural network that is specific to the first machine learning task, the path specifying, for each of the layers, a proper subset of the modular neural networks in the layer that are designated as active when performing the first machine learning task; and causing the super neural network to process the first network input using (i) for each layer, the modular neural networks in the layer that are designated as active by the selected path and (ii) the set of one or more output layers corresponding to the identified first machine learning task.Type: ApplicationFiled: October 16, 2023Publication date: February 8, 2024Inventors: Daniel Pieter Wierstra, Chrisantha Thomas Fernando, Alexander Pritzel, Dylan Sunil Banarse, Charles Blundell, Andrei-Alexandru Rusu, Yori Zwols, David Ha
-
Patent number: 11790238Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using multi-task neural networks. One of the methods includes receiving a first network input and data identifying a first machine learning task to be performed on the first network input; selecting a path through the plurality of layers in a super neural network that is specific to the first machine learning task, the path specifying, for each of the layers, a proper subset of the modular neural networks in the layer that are designated as active when performing the first machine learning task; and causing the super neural network to process the first network input using (i) for each layer, the modular neural networks in the layer that are designated as active by the selected path and (ii) the set of one or more output layers corresponding to the identified first machine learning task.Type: GrantFiled: August 17, 2020Date of Patent: October 17, 2023Assignee: DeepMind Technologies LimitedInventors: Daniel Pieter Wierstra, Chrisantha Thomas Fernando, Alexander Pritzel, Dylan Sunil Banarse, Charles Blundell, Andrei-Alexandru Rusu, Yori Zwols, David Ha
-
Patent number: 11164066Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing inputs using recurrent neural networks. One of the systems includes a main recurrent neural network comprising one or more recurrent neural network layers and a respective hyper recurrent neural network corresponding to each of the one or more recurrent neural network layers, wherein each hyper recurrent neural network is configured to, at each of a plurality of time steps: process the layer input at the time step to the corresponding recurrent neural network layer, the current layer hidden state of the corresponding recurrent neural network layer, and a current hypernetwork hidden state of the hyper recurrent neural network to generate an updated hypernetwork hidden state.Type: GrantFiled: September 26, 2017Date of Patent: November 2, 2021Assignee: Google LLCInventors: Andrew M. Dai, Quoc V. Le, David Ha
-
Publication number: 20200380372Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using multi-task neural networks. One of the methods includes receiving a first network input and data identifying a first machine learning task to be performed on the first network input; selecting a path through the plurality of layers in a super neural network that is specific to the first machine learning task, the path specifying, for each of the layers, a proper subset of the modular neural networks in the layer that are designated as active when performing the first machine learning task; and causing the super neural network to process the first network input using (i) for each layer, the modular neural networks in the layer that are designated as active by the selected path and (ii) the set of one or more output layers corresponding to the identified first machine learning task.Type: ApplicationFiled: August 17, 2020Publication date: December 3, 2020Inventors: Daniel Pieter Wierstra, Chrisantha Thomas Fernando, Alexander Pritzel, Dylan Sunil Banarse, Charles Blundell, Andrei-Alexandru Rusu, Yori Zwols, David Ha
-
Patent number: 10748065Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using multi-task neural networks. One of the methods includes receiving a first network input and data identifying a first machine learning task to be performed on the first network input; selecting a path through the plurality of layers in a super neural network that is specific to the first machine learning task, the path specifying, for each of the layers, a proper subset of the modular neural networks in the layer that are designated as active when performing the first machine learning task; and causing the super neural network to process the first network input using (i) for each layer, the modular neural networks in the layer that are designated as active by the selected path and (ii) the set of one or more output layers corresponding to the identified first machine learning task.Type: GrantFiled: July 30, 2019Date of Patent: August 18, 2020Assignee: DeepMind Technologies LimitedInventors: Daniel Pieter Wierstra, Chrisantha Thomas Fernando, Alexander Pritzel, Dylan Sunil Banarse, Charles Blundell, Andrei-Alexandru Rusu, Yori Zwols, David Ha
-
Publication number: 20200004933Abstract: A method, apparatus, article of manufacture, and a memory structure for providing a security infrastructure that permits the programming of limited hardware resources that can accept newly downloaded applications and securely support a very large number of services offered by content providers each have the potential to utilize their own independent CAS/DRM system. The CE device owner can consume content from a variety of sources and enable switching among different and existing CAS/DRM security profiles as required by the content provider applications loaded in CE devices.Type: ApplicationFiled: July 8, 2019Publication date: January 2, 2020Applicant: Inside SecureInventors: Ronald P. Cocchi, Michael A. Gorman, Jacob T. Carson, Matthew A. Skubiszewski, David Ha
-
Publication number: 20190354868Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using multi-task neural networks. One of the methods includes receiving a first network input and data identifying a first machine learning task to be performed on the first network input; selecting a path through the plurality of layers in a super neural network that is specific to the first machine learning task, the path specifying, for each of the layers, a proper subset of the modular neural networks in the layer that are designated as active when performing the first machine learning task; and causing the super neural network to process the first network input using (i) for each layer, the modular neural networks in the layer that are designated as active by the selected path and (ii) the set of one or more output layers corresponding to the identified first machine learning task.Type: ApplicationFiled: July 30, 2019Publication date: November 21, 2019Inventors: Daniel Pieter Wierstra, Chrisantha Thomas Fernando, Alexander Pritzel, Dylan Sunil Banarse, Charles Blundell, Andrei-Alexandru Rusu, Yori Zwols, David Ha
-
Patent number: 10348501Abstract: A method, apparatus, article of manufacture, and a memory structure for providing a security infrastructure that permits the programming of limited hardware resources that can accept newly downloaded applications and securely support a very large number of services offered by content providers each have the potential to utilize their own independent CAS/DRM system. The CE device owner can consume content from a variety of sources and enable switching among different and existing CAS/DRM security profiles as required by the content provider applications loaded in CE devices.Type: GrantFiled: July 11, 2016Date of Patent: July 9, 2019Assignee: Inside SecureInventors: Ronald P. Cocchi, Michael A. Gorman, Jacob T. Carson, Matthew A. Skubiszewski, David Ha
-
Publication number: 20170012952Abstract: A method, apparatus, article of manufacture, and a memory structure for providing a security infrastructure that permits the programming of limited hardware resources that can accept newly downloaded applications and securely support a very large number of services offered by content providers each have the potential to utilize their own independent CAS/DRM system. The CE device owner can consume content from a variety of sources and enable switching among different and existing CAS/DRM security profiles as required by the content provider applications loaded in CE devices.Type: ApplicationFiled: July 11, 2016Publication date: January 12, 2017Inventors: Ronald P. Cocchi, Michael A. Gorman, Jacob T. Carson, Matthew A. Skubiszewski, David Ha
-
Patent number: 8618211Abstract: A thermally conductive grease includes 2 vol % to 15 vol % of a combination of three polyorganosiloxanes and 65 vol % to 98 vol % of a thermally conductive filler. The grease may be used as a thermal interface material for dissipating heat from (opto)electronic devices, in both TIM1 and TIM2 applications.Type: GrantFiled: January 22, 2010Date of Patent: December 31, 2013Assignee: Dow Corning CorporationInventors: Dorab Edul Bhagwagar, David Ha, Sarah Nicol, Kai Su
-
Publication number: 20110272119Abstract: A thermally conductive grease includes 2 vol % to 15 vol % of a combination of three polyorganosiloxanes and 65 vol % to 98 vol % of a thermally conductive filler. The grease may be used as a thermal interface material for dissipating heat from (opto)electronic devices, in both TIM1 and TIM2 applications.Type: ApplicationFiled: January 22, 2010Publication date: November 10, 2011Inventors: Dorab Edul Bhagwagar, David Ha, Sarah Nicol, Kai Su
-
Super encrypted storage and retrieval of media programs in a hard-paired receiver and storage device
Publication number: 20070133795Abstract: A method and apparatus for storing and retrieving program material for subsequent replay is disclosed.Type: ApplicationFiled: February 2, 2007Publication date: June 14, 2007Inventors: Raynold Kahn, Gregory Gagnon, David Ha, Peter Klauss, Christopher Curren, Thomas James -
Publication number: 20070118770Abstract: A method and apparatus for storing and retrieving program material for subsequent replay is disclosed. The method comprises the steps of receiving a data stream comprising the program material encrypted according to a first encryption key, decrypting the program material; re-encrypting the program material according to a second encryption key; and storing the re-encrypted material in a media storage device. The program material is played back by retrieving the re-encrypted material from the media storage device and decrypting the re-encrypted program material. In one embodiment, the second encryption key is derived from metadata describing replay rights. In a further embodiment, the media storage device also stores the second encryption key which has been further encrypted by a key that is unique to the device used to receive the program material.Type: ApplicationFiled: January 18, 2007Publication date: May 24, 2007Inventors: Raynold Kahn, Gregory Gagnon, David Ha, Peter Klauss, Christopher Curren, Thomas James
-
Publication number: 20060161976Abstract: An apparatus and method for transmission security is presented. A receiver communicatively coupleable to an identifiable conditional access module (CAM) to enable reception of the service uses an authenticator to control the operability of the receiver according to a comparison between the identified conditional access module and a list of conditional access modules.Type: ApplicationFiled: February 10, 2006Publication date: July 20, 2006Inventors: Raynold Kahn, Gregory Gagnon, David Ha, Dennis Flaharty
-
Publication number: 20050249350Abstract: A method, apparatus, and article of manufacture provide the ability to process digital media content. A secret key (Ka) is obtained and encrypted into an authorization key message (AKM) message. One or more entitlement keys (Ke) are then obtained and encrypted using the secret key (Ka) as an encryption key into an entitlement management message (EMM). One or more media content encryption keys (Kc) are obtained and encrypted using one or more of the entitlement keys (Ke) as encryption keys into an entitlement control message (ECM). Digital media data content (Dc) is then obtained and encrypted using the one or more media content encryption keys (Kc). The AKM, EMM, ECM, and encrypted digital media data content are then delivered to a set-top box that is configured to provide digital media services with the content.Type: ApplicationFiled: May 2, 2005Publication date: November 10, 2005Inventors: Raynold Kahn, David Ha, Thomas James, Peter Klauss
-
Patent number: 6924346Abstract: Silicone resins comprising 5 to 50 mole % of (PhSiO3-x)/2(OH)x) units and 50 to 95 mole % (HSiO(3-x)/2(OH)x), where Ph is a phenyl group, x has a value of 0, 1 or 2 and wherein the cured silicone resin has a critical surface free energy of 30 dynes/cm or higher. These resins are useful as etch stop layers for organic dielectric materials having a critical surface free energy of 40 dynes/cm or higher.Type: GrantFiled: October 4, 2002Date of Patent: August 2, 2005Assignee: Dow Corning CorporationInventors: Ronald Boisvert, Stelian Grigoras, David Ha, Brian Harkness, Craig Yeakle
-
Publication number: 20040186223Abstract: Silicone resins comprising 5 to 50 mole % of (PhSiO3−x)/2(OH)x) units and 50 to 95 mole % (HSiO(3−x)/2(OH)x), where Ph is a phenyl group, x has a value of 0, 1 or 2 and wherein the cured silicone resin has a critical surface free energy of 30 dynes/cm or higher. These resins are useful as etch stop layers for organic dielectric materials having a critical surface free energy of 40 dynes/cm or higher.Type: ApplicationFiled: April 1, 2004Publication date: September 23, 2004Inventors: Ronald Boisvert, Stelian Grigoras, David Ha, Brian Harkness, Craig Yeakle