Accelerating Reinforcement Learning Agent with EEG-based Implicit Human Feedback

TL;DR

This paper introduces a novel method for accelerating reinforcement learning by using EEG-based implicit human feedback, specifically error-related potentials, which reduces human burden and enhances learning efficiency in complex environments.

Contribution

It presents a new RL framework that integrates EEG-derived implicit feedback, enabling zero-shot ErrP learning transfer across games and scaling to complex environments.

Findings

ErrPs can be learned zero-shot and transferred across games

Implicit feedback improves RL label efficiency and robustness

Real user experiments show accelerated learning in complex environments

Abstract

Providing Reinforcement Learning (RL) agents with human feedback can dramatically improve various aspects of learning. However, previous methods require human observer to give inputs explicitly (e.g., press buttons, voice interface), burdening the human in the loop of RL agent's learning process. Further, it is sometimes difficult or impossible to obtain the explicit human advise (feedback), e.g., autonomous driving, disabled rehabilitation, etc. In this work, we investigate capturing human's intrinsic reactions as implicit (and natural) feedback through EEG in the form of error-related potentials (ErrP), providing a natural and direct way for humans to improve the RL agent learning. As such, the human intelligence can be integrated via implicit feedback with RL algorithms to accelerate the learning of RL agent. We develop three reasonably complex 2D discrete navigational games to experimentally evaluate the overall performance of the proposed work. Major contributions of our work are as follows, (i) we propose and experimentally validate the zero-shot learning of ErrPs, where the ErrPs can be learned for one game, and transferred to other unseen games, (ii) we propose a novel RL framework for integrating implicit human feedbacks via ErrPs with RL agent, improving the label efficiency and robustness to human mistakes, and (iii) compared to prior works, we scale the application of ErrPs to reasonably complex environments, and demonstrate the significance of our approach for accelerated learning through real user experiments.