EEGEncoder: Advancing BCI with Transformer-Based Motor Imagery Classification

TL;DR

EEGEncoder is a novel transformer-based deep learning framework that significantly improves EEG motor imagery classification accuracy by capturing temporal and spatial features more effectively than previous methods.

Contribution

The paper introduces EEGEncoder with a dual-stream fusion architecture and parallel structures, advancing BCI motor imagery classification beyond existing techniques.

Findings

Outperforms state-of-the-art methods on BCI Competition IV-2a dataset

Effectively captures temporal and spatial EEG features

Enhances classification accuracy with novel fusion architecture

Abstract

Brain-computer interfaces (BCIs) harness electroencephalographic signals for direct neural control of devices, offering a significant benefit for individuals with motor impairments. Traditional machine learning methods for EEG-based motor imagery (MI) classification encounter challenges such as manual feature extraction and susceptibility to noise.This paper introduces EEGEncoder, a deep learning framework that employs modified transformers and TCNs to surmount these limitations. We innovatively propose a fusion architecture, namely Dual-Stream Temporal-Spatial Block (DSTS), to capture temporal and spatial features, improving the accuracy of Motor Imagery classification task. Additionally, we use multiple parallel structures to enhance the performance of the model. When tested on the BCI Competition IV-2a dataset, our model results outperform current state-of-the-art techniques.