RMETNet: Structural Details for a Motor Imagery EEG Classification Model

RMETNet is a dataset containing structural details for a novel deep learning framework designed for motor imagery electroencephalogram (MI-EEG) analysis. The framework integrates spatio-temporal convolution and Riemannian geometry features, and was tested on the BCI Competition IV 2a and 2b datasets. The dataset, authored by Yun Zhao and last updated on 2026-04-22, is a 9.5 KB XLS file.

Use Cases

• Replicating the RMETNet model's architecture based on the provided structural details.
• Studying domain adaptation techniques for EEG data using the described Maximum Mean Discrepancy (MMD) loss.
• Analyzing the integration of Riemannian geometry features for handling inter-subject variability in EEG signals.
• Benchmarking new EEG classification models against the performance metrics reported for RMETNet.

Strengths

• Model performance is quantified with specific cross-subject and subject-dependent accuracy scores on two benchmark datasets.
• The dataset is associated with a publicly available code repository, facilitating reproducibility.
• It is released under a permissive CC-BY-4.0 license.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small at 9.5 KB, indicating it contains metadata or model parameters rather than raw EEG data.

Provenance

Source

figshare, authored by Yun Zhao.

Collection Method

Likely contains parameters, architecture details, or results from the RMETNet model.

Time Range

The dataset itself was updated in 2026; the underlying research uses the BCI Competition IV datasets.

Freshness

Last updated 2026-04-22 17:35:05; freshness should be verified.

Geography

Geographic coverage of the source EEG data is not specified.