moabb

A research repository of deep learning on electroencephalographic (EEG) for Motor imagery(MI), including eeg data processing(visualization & analysis), papers(research and summary), deep learning models(reproduction and experiments).

The Neural Interface Foundation Challenge — NeurIPS 2026 Sydney. An open-source brain-decoding benchmark with four tracks across EEG, EMG, sleep, and BCI.

Processing EEG data using Speechbrain-MOABB and model tuning to get best results

Motor-imagery EEG decoding with covariance alignment (EA/Riemannian) and information-maximization test-time adaptation on an EEGNet backbone — benchmarked on BCI Competition IV-2a & IV-2b (within-subject and cross-subject).

Honest dataset cards for public EEG/BCI datasets: leakage-aware baselines, chance/significance, per-class metrics, and MOABB + EEGDash metadata. Not medical software.

EEG classification project. Uses a variety of classifiers and methods to parse EEG data.

This example compares the classification performance of linear support vector machine (LinearSVC) on the Riemannian Transfer Learning method (RPA, Rodrigues et al., 2018) and the golden-standard subject-wise train-test cross-validation method using real P300 BCI data.

Pipeline d'Adaptation de Domaine Non-Supervisée pour Signaux EEG

Channel selection for P300 BCI using a continuous-weight evolutionary strategy. Evaluated against oracle, uniform, and SNR-based baselines on the Lee2019_ERP dataset, with cross-subject transfer analysis.

NeuroKiln（脑炉）— 协议优先、可复现的 BCI 算法评测与调参平台

P300 speller EEG classification: xDAWN+LDA baseline vs EEGNet (PyTorch), with LOSO cross-validation revealing the cross-subject transfer gap and a fine-tuning calibration strategy

EEG Classification using MiniRocket + Ridge Classifier on BCI Competition IV Dataset 2a. Reproducible pipeline for Time-Series feature extraction and classification.

Cross-session EEG motor-imagery classification with domain adaptation: four transfer pipelines (MAP/DWP/MMP/BDP) over a configurable method bank, tested on public MI datasets.

Reproducible benchmark of motor-imagery EEG decoder robustness to channel loss, reduced montages, and cross-session transfer.

EEG-based asynchronous BCI control vs non-control classification pipeline using CSP and machine learning.

Public cleanroom projection of the RIEC-BCI evaluation and reproducibility workflow.

cajd: Composite Approximate Joint Diagonalization in Python — EEG/ERP source separation mixing linear (covariance) and bilinear (spatio-temporal) models. scikit-learn, MNE, MOABB and pyRiemann compatible.