Technology Category
Machine Learning: Learning with ManifoldsCognitive Modeling & Cognitive Systems: Neural Spike CodingComputer Vision: Multi-modal Vision
LGL-BCI: A Motor-Imagery-Based Brain-Computer Interface with Geometric Learning
🤖 AI Summary
To address the challenges of high variability in EEG amplitude/phase, complex spatial correlations, and difficulties in model lightweighting and real-time inference for motor imagery brain–computer interfaces (MI-BCIs), this paper proposes a lightweight geometric deep learning architecture operating on the symmetric positive definite (SPD) manifold. We innovatively design a feature-decomposition-based adaptive channel selection module and introduce a lossless tensor transformation acceleration mechanism. By integrating a lightweight CNN–GNN hybrid structure, our method enables efficient geometric modeling directly on the SPD manifold. Evaluated on two devices across two benchmark datasets, our approach achieves a mean classification accuracy of 82.54%—a 20.32% improvement over the state-of-the-art—while reducing model parameters to only 64.9K (a 64.7% reduction). The architecture thus simultaneously advances accuracy, computational efficiency, and practical deployability for real-time MI-BCI applications.
Technology Category
Application Category
📝 Abstract
Brain--computer interfaces are groundbreaking technology whereby brain signals are used to control external devices. Despite some advances in recent years, electroencephalogram (EEG)-based motor-imagery tasks face challenges, such as amplitude and phase variability and complex spatial correlations, with a need for smaller models and faster inference. In this study, we develop a prototype, called the Lightweight Geometric Learning Brain--Computer Interface (LGL-BCI), which uses our customized geometric deep learning architecture for swift model inference without sacrificing accuracy. LGL-BCI contains an EEG channel selection module via a feature decomposition algorithm to reduce the dimensionality of a symmetric positive definite matrix, providing adaptiveness among the continuously changing EEG signal. Meanwhile, a built-in lossless transformation helps boost the inference speed. The performance of our solution was evaluated using two real-world EEG devices and two public EEG datasets. LGL-BCI demonstrated significant improvements, achieving an accuracy of 82.54% compared to 62.22% for the state-of-the-art approach. Furthermore, LGL-BCI uses fewer parameters (64.9K vs. 183.7K), highlighting its computational efficiency. These findings underscore both the superior accuracy and computational efficiency of LGL-BCI, demonstrating the feasibility and robustness of geometric deep learning in motor-imagery brain--computer interface applications.
Problem
Research questions and friction points this paper is trying to address.
Brain-Computer Interface
EEG Signal Processing
Model Efficiency
Innovation
Methods, ideas, or system contributions that make the work stand out.
LGL-BCI
Geometric Deep Learning
Efficient Resource Utilization
🔎 Similar Papers
No similar papers found.
