Patents by Inventor Matthew William Hoffman

Matthew William Hoffman 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).

  • TRAINING MACHINE LEARNING MODELS BY DETERMINING UPDATE RULES USING NEURAL NETWORKS
    Publication number: 20250335771
    Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for training machine learning models. One method includes obtaining a machine learning model, wherein the machine learning model comprises one or more model parameters, and the machine learning model is trained using gradient descent techniques to optimize an objective function; determining an update rule for the model parameters using a recurrent neural network (RNN); and applying a determined update rule for a final time step in a sequence of multiple time steps to the model parameters.
    Type: Application
    Filed: March 11, 2025
    Publication date: October 30, 2025
    Inventors: Misha Man Ray Denil, Tom Schaul, Marcin Andrychowicz, Joao Ferdinando Gomes de Freitas, Sergio Gomez Colmenarejo, Matthew William Hoffman, David Benjamin Pfau
  • Training machine learning models by determining update rules using neural networks
    Patent number: 12271823
    Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for training machine learning models. One method includes obtaining a machine learning model, wherein the machine learning model comprises one or more model parameters, and the machine learning model is trained using gradient descent techniques to optimize an objective function; determining an update rule for the model parameters using a recurrent neural network (RNN); and applying a determined update rule for a final time step in a sequence of multiple time steps to the model parameters.
    Type: Grant
    Filed: March 8, 2023
    Date of Patent: April 8, 2025
    Assignee: DeepMind Technologies Limited
    Inventors: Misha Man Ray Denil, Tom Schaul, Marcin Andrychowicz, Joao Ferdinando Gomes de Freitas, Sergio Gomez Colmenarejo, Matthew William Hoffman, David Benjamin Pfau
  • Distributional reinforcement learning for continuous control tasks
    Patent number: 11948085
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Grant
    Filed: April 19, 2023
    Date of Patent: April 2, 2024
    Assignee: DeepMind Technologies Limited
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • DISTRIBUTIONAL REINFORCEMENT LEARNING FOR CONTINUOUS CONTROL TASKS
    Publication number: 20230409907
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Application
    Filed: April 19, 2023
    Publication date: December 21, 2023
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • TRAINING MACHINE LEARNING MODELS BY DETERMINING UPDATE RULES USING NEURAL NETWORKS
    Publication number: 20230376771
    Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for training machine learning models. One method includes obtaining a machine learning model, wherein the machine learning model comprises one or more model parameters, and the machine learning model is trained using gradient descent techniques to optimize an objective function; determining an update rule for the model parameters using a recurrent neural network (RNN); and applying a determined update rule for a final time step in a sequence of multiple time steps to the model parameters.
    Type: Application
    Filed: March 8, 2023
    Publication date: November 23, 2023
    Inventors: Misha Man Ray Denil, Tom Schaul, Marcin Andrychowicz, Joao Ferdinando Gomes de Freitas, Sergio Gomez Colmenarejo, Matthew William Hoffman, David Benjamin Pfau
  • Distributional reinforcement learning for continuous control tasks
    Patent number: 11663475
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Grant
    Filed: September 15, 2022
    Date of Patent: May 30, 2023
    Assignee: DeepMind Technologies Limited
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • Training machine learning models by determining update rules using recurrent neural networks
    Patent number: 11615310
    Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for training machine learning models. One method includes obtaining a machine learning model, wherein the machine learning model comprises one or more model parameters, and the machine learning model is trained using gradient descent techniques to optimize an objective function; determining an update rule for the model parameters using a recurrent neural network (RNN); and applying a determined update rule for a final time step in a sequence of multiple time steps to the model parameters.
    Type: Grant
    Filed: May 19, 2017
    Date of Patent: March 28, 2023
    Assignee: DeepMind Technologies Limited
    Inventors: Misha Man Ray Denil, Tom Schaul, Marcin Andrychowicz, Joao Ferdinando Gomes de Freitas, Sergio Gomez Colmenarejo, Matthew William Hoffman, David Benjamin Pfau
  • DISTRIBUTIONAL REINFORCEMENT LEARNING FOR CONTINUOUS CONTROL TASKS
    Publication number: 20230020071
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Application
    Filed: September 15, 2022
    Publication date: January 19, 2023
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • Distributional reinforcement learning for continuous control tasks
    Patent number: 11481629
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Grant
    Filed: October 29, 2018
    Date of Patent: October 25, 2022
    Assignee: DeepMind Technologies Limited
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • DISTRIBUTIONAL REINFORCEMENT LEARNING FOR CONTINUOUS CONTROL TASKS
    Publication number: 20200293883
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network that is used to select actions to be performed by a reinforcement learning agent interacting with an environment. In particular, the actions are selected from a continuous action space and the system trains the action selection neural network jointly with a distribution Q network that is used to update the parameters of the action selection neural network.
    Type: Application
    Filed: October 29, 2018
    Publication date: September 17, 2020
    Inventors: David Budden, Matthew William Hoffman, Gabriel Barth-Maron
  • TRAINING MACHINE LEARNING MODELS
    Publication number: 20190220748
    Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for training machine learning models. One method includes obtaining a machine learning model, wherein the machine learning model comprises one or more model parameters, and the machine learning model is trained using gradient descent techniques to optimize an objective function; determining an update rule for the model parameters using a recurrent neural network (RNN); and applying a determined update rule for a final time step in a sequence of multiple time steps to the model parameters.
    Type: Application
    Filed: May 19, 2017
    Publication date: July 18, 2019
    Inventors: Misha Man Ray Denil, Tom Schaul, Marcin Andrychowicz, Joao Ferdinando Gomes de Freitas, Sergio Gomez Colmenarejo, Matthew William Hoffman, David Benjamin Pfau