R Package: DA4BCI
Published:
Package: DA4BCI (Data Augmentation for Brain-Computer Interface)
Title: A Unified Framework for Domain Adaptation in EEG-based BCI
Version: 0.1.0
Authors: Yiming Shen, David Degras
License: MIT + file LICENSE
Overview
DA4BCI provides a unified interface for domain adaptation in EEG-based brain-computer interface workflows. The package focuses on aligning source and target EEG distributions across sessions or subjects to reduce shift and improve model robustness.
Methods Included (DESCRIPTION)
- Subspace/feature adaptation: TCA, SA, MIDA, CORAL.
- Geometric/manifold transport: GFK, ART, PT, M3D, OT.
- Evaluation metrics: Maximum Mean Discrepancy (MMD), Wasserstein Distance, Energy Distance.
Dependency Snapshot
- Depends:
R (>= 3.5.0),ggplot2,Rtsne,MASS,RSpectra,geigen,pracma. - Imports:
transport (>= 0.13-1),gridExtra, plus core dependencies above.
Installation & Usage
You can install the development version from GitHub:
# install.packages("devtools")
devtools::install_github("Yiming-S/DA4BCI")
Example: Generating Artificial Covariance Matrices
library(DA4BCI)
# Load sample EEG covariance data
data("sample_covariances")
# Generate 100 artificial trials using Riemannian interpolation
synthetic_data <- generate_riemannian_data(cov_matrices = sample_covariances,
n_samples = 100,
method = "geodesic")