Patents by Inventor Alexander Vezhnevets

Alexander Vezhnevets 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).

  • ACTION SELECTION FOR REINFORCEMENT LEARNING USING A MANAGER NEURAL NETWORK THAT GENERATES GOAL VECTORS DEFINING AGENT OBJECTIVES
    Publication number: 20230090824
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to select actions to be performed by an agent that interacts with an environment. The system comprises a manager neural network subsystem and a worker neural network subsystem. The manager subsystem is configured to, at each of the multiple time steps, generate a final goal vector for the time step. The worker subsystem is configured to, at each of multiple time steps, use the final goal vector generated by the manager subsystem to generate a respective action score for each action in a predetermined set of actions.
    Type: Application
    Filed: November 30, 2022
    Publication date: March 23, 2023
    Inventors: Simon Osindero, Koray Kavukcuoglu, Alexander Vezhnevets
  • Action selection for reinforcement learning using a manager neural network that generates goal vectors defining agent objectives
    Patent number: 11537887
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to select actions to be performed by an agent that interacts with an environment. The system comprises a manager neural network subsystem and a worker neural network subsystem. The manager subsystem is configured to, at each of the multiple time steps, generate a final goal vector for the time step. The worker subsystem is configured to, at each of multiple time steps, use the final goal vector generated by the manager subsystem to generate a respective action score for each action in a predetermined set of actions.
    Type: Grant
    Filed: May 5, 2020
    Date of Patent: December 27, 2022
    Assignee: DeepMind Technologies Limited
    Inventors: Simon Osindero, Koray Kavukcuoglu, Alexander Vezhnevets
  • ACTION SELECTION FOR REINFORCEMENT LEARNING USING NEURAL NETWORKS
    Publication number: 20200265313
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to select actions to be performed by an agent that interacts with an environment. The system comprises a manager neural network subsystem and a worker neural network subsystem. The manager subsystem is configured to, at each of the multiple time steps, generate a final goal vector for the time step. The worker subsystem is configured to, at each of multiple time steps, use the final goal vector generated by the manager subsystem to generate a respective action score for each action in a predetermined set of actions.
    Type: Application
    Filed: May 5, 2020
    Publication date: August 20, 2020
    Inventors: Simon Osindero, Koray Kavukcuoglu, Alexander Vezhnevets
  • Action selection for reinforcement learning using neural networks
    Patent number: 10679126
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to select actions to be performed by an agent that interacts with an environment. The system comprises a manager neural network subsystem and a worker neural network subsystem. The manager subsystem is configured to, at each of the multiple time steps, generate a final goal vector for the time step. The worker subsystem is configured to, at each of multiple time steps, use the final goal vector generated by the manager subsystem to generate a respective action score for each action in a predetermined set of actions.
    Type: Grant
    Filed: July 15, 2019
    Date of Patent: June 9, 2020
    Assignee: DeepMind Technologies Limited
    Inventors: Simon Osindero, Koray Kavukcuoglu, Alexander Vezhnevets
  • ACTION SELECTION FOR REINFORCEMENT LEARNING USING NEURAL NETWORKS
    Publication number: 20190340509
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to select actions to be performed by an agent that interacts with an environment. The system comprises a manager neural network subsystem and a worker neural network subsystem. The manager subsystem is configured to, at each of the multiple time steps, generate a final goal vector for the time step. The worker subsystem is configured to, at each of multiple time steps, use the final goal vector generated by the manager subsystem to generate a respective action score for each action in a predetermined set of actions.
    Type: Application
    Filed: July 15, 2019
    Publication date: November 7, 2019
    Inventors: Simon Osindero, Koray Kavukcuoglu, Alexander Vezhnevets