Bernardo Avila Pires

Title: Bernardo Avila Pires - Innovator in Reinforcement Learning

Introduction

Bernardo Avila Pires is an accomplished inventor based in London, GB. He is currently associated with DeepMind Technologies Limited, where he contributes to advancements in artificial intelligence and reinforcement learning. Although he has not yet been granted any patents, his work is significant in the field of machine learning.

Latest Patent Applications

Bernardo has been involved in several innovative patent applications. One of his latest applications is titled "Exploration by Bootstrapped Prediction." This application proposes an iterative method to train an action selection system of a reinforcement learning system, based on a reward function that defines a reward value for each action. The reward value includes an intrinsic reward term generated based on the outputs of two encoder models: an online encoder model and a target encoder model. The online encoder model is iteratively trained based on a loss function, while the target encoder model is updated to align more closely with the online encoder model.

Another notable application is "Hierarchical Reinforcement Learning at Scale." This invention describes a system and method for controlling an agent interacting with an environment to perform a task. The method involves receiving an observation characterizing a state of the environment at various time steps. It also includes determining a goal representation for the agent to achieve and processing the observation and goal representation using a low-level controller neural network to generate a low-level policy output that defines the action to be performed by the agent.

Conclusion

Bernardo Avila Pires is making significant contributions to the field of reinforcement learning through his innovative patent applications. His work at DeepMind Technologies Limited showcases his commitment to advancing artificial intelligence technologies.