Patents by Inventor Vlad Firoiu

Vlad Firoiu 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).

  • Distributed training using actor-critic reinforcement learning with off-policy correction factors
    Patent number: 12299574
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Grant
    Filed: October 16, 2023
    Date of Patent: May 13, 2025
    Assignee: DeepMind Technologies Limited
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning
  • DISTRIBUTED TRAINING USING ACTOR-CRITIC REINFORCEMENT LEARNING WITH OFF-POLICY CORRECTION FACTORS
    Publication number: 20240127060
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Application
    Filed: October 16, 2023
    Publication date: April 18, 2024
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning
  • Distributed training using actor-critic reinforcement learning with off-policy correction factors
    Patent number: 11868894
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Grant
    Filed: January 4, 2023
    Date of Patent: January 9, 2024
    Assignee: DeepMind Technologies Limited
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning
  • DISTRIBUTED TRAINING USING ACTOR-CRITIC REINFORCEMENT LEARNING WITH OFF-POLICY CORRECTION FACTORS
    Publication number: 20230153617
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Application
    Filed: January 4, 2023
    Publication date: May 18, 2023
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning
  • Distributed training using actor-critic reinforcement learning with off-policy correction factors
    Patent number: 11593646
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Grant
    Filed: February 5, 2019
    Date of Patent: February 28, 2023
    Assignee: DeepMind Technologies Limited
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning
  • DISTRIBUTED TRAINING USING ACTOR-CRITIC REINFORCEMENT LEARNING WITH OFF-POLICY CORRECTION FACTORS
    Publication number: 20210034970
    Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network used to select actions to be performed by an agent interacting with an environment. In one aspect, a system comprises a plurality of actor computing units and a plurality of learner computing units. The actor computing units generate experience tuple trajectories that are used by the learner computing units to update learner action selection neural network parameters using a reinforcement learning technique. The reinforcement learning technique may be an off-policy actor critic reinforcement learning technique.
    Type: Application
    Filed: February 5, 2019
    Publication date: February 4, 2021
    Inventors: Hubert Josef Soyer, Lasse Espeholt, Karen Simonyan, Yotam Doron, Vlad Firoiu, Volodymyr Mnih, Koray Kavukcuoglu, Remi Munos, Thomas Ward, Timothy James Alexander Harley, Iain Robert Dunning