RLEEGNet: Integrating Brain-Computer Interfaces with Adaptive AI for Intuitive Responsiveness and High-Accuracy Motor Imagery Classification

TL;DR

RLEEGNet tackles the challenge of real-time, intuitive motor imagery Brain-Computer Interface (BCI) control by combining reinforcement learning with a Deep Q-Network (DQN) framework and a 1D-CNN-LSTM Online Q-Network. It introduces CSP-based One-Vs.-Rest (OVR) preprocessing and a novel CSP Space transformation to preserve temporal structure while extracting discriminative spatial features, coupled with Welch PSD and other statistical features. The approach is evaluated on two MI-EEG datasets (GigaScience and BCI-IV2a), achieving up to 100% reward-based accuracy and high intra-session accuracies (average around 96.7% on BCI-IV2a), demonstrating strong adaptability and robustness. The work highlights the significance of carefully designed reward signals, time-window selection, and feature extraction in enabling adaptive, high-precision BCI systems suitable for diverse cognitive states and motor intentions.

Abstract

Current approaches to prosthetic control are limited by their reliance on traditional methods, which lack real-time adaptability and intuitive responsiveness. These limitations are particularly pronounced in assistive technologies designed for individuals with diverse cognitive states and motor intentions. In this paper, we introduce a framework that leverages Reinforcement Learning (RL) with Deep Q-Networks (DQN) for classification tasks. Additionally, we present a preprocessing technique using the Common Spatial Pattern (CSP) for multiclass motor imagery (MI) classification in a One-Versus-The-Rest (OVR) manner. The subsequent 'csp space' transformation retains the temporal dimension of EEG signals, crucial for extracting discriminative features. The integration of DQN with a 1D-CNN-LSTM architecture optimizes the decision-making process in real-time, thereby enhancing the system's adaptability to the user's evolving needs and intentions. We elaborate on the data processing methods for two EEG motor imagery datasets. Our innovative model, RLEEGNet, incorporates a 1D-CNN-LSTM architecture as the Online Q-Network within the DQN, facilitating continuous adaptation and optimization of control strategies through feedback. This mechanism allows the system to learn optimal actions through trial and error, progressively improving its performance. RLEEGNet demonstrates high accuracy in classifying MI-EEG signals, achieving as high as 100% accuracy in MI tasks across both the GigaScience (3-class) and BCI-IV-2a (4-class) datasets. These results highlight the potential of combining DQN with a 1D-CNN-LSTM architecture to significantly enhance the adaptability and responsiveness of BCI systems.

Figures (9)

• Figure 1: EEG Waveforms & Indication of Mental State
• Figure :
• Figure :
• Figure :
• Figure :