ADFCNN-MI

KO | EN

gitlite — search

#python #java #python3 #arduino #golang #machine-learning #rust #html #flask #javascript #seismology #nodejs

★ 3 Open GitHub ↗

Motor Imagery Classification using Adaptive Deep Feature Learning

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# ADFCNN-MI: Motor Imagery Classification using Adaptive Deep Feature Learning

This repository contains the implementation of ADFCNN (Adaptive Deep Feature Convolutional Neural Network) for motor imagery (MI) classification using EEG signals from the BCI Competition IV 2a dataset.

## Project Structure
```
ADFCNN-MI/
├── ADFCNN/
│   ├── configs/
│   │   └── bcicompet2a_config.yaml
│   ├── dataloader/
│   │   ├── augmentation.py
│   │   └── bci_compet.py
│   ├── model/
│   │   ├── ADFCNN.py
│   │   └── litmodel.py
│   ├── utils/
│   │   ├── setup_utils.py
│   │   └── training_utils.py
│   ├── training.py
│   ├── evaluation.py
│   └── train_all_subjects.py
├── requirements.txt
└── README.md
```

## Features
- Implementation of ADFCNN model for MI classification
- Support for BCI Competition IV 2a dataset
- Cross-session evaluation
- Data augmentation techniques
- Comprehensive evaluation metrics

## Performance
- Average Accuracy: 71.03%
- Average Kappa Score: 0.6137
- Per-Class Performance:
  - Left Hand: 75.00%
  - Right Hand: 73.15%
  - Feet: 64.35%
  - Tongue: 71.60%

## Requirements
- Python 3.8+
- PyTorch
- MNE
- Braindecode
- PyTorch Lightning
- scikit-learn
- Other dependencies listed in requirements.txt

## Installation
1. Clone the repository:
```bash
git clone https://github.com/yourusername/ADFCNN-MI.git
cd ADFCNN-MI
```

2. Install dependencies:
```bash
pip install -r requirements.txt
```

3. Configure data paths in `ADFCNN/configs/bcicompet2a_config.yaml`

## Usage
1. Training a single subject:
```bash
python ADFCNN/training.py --config_name bcicompet2a_config --gpu_num 0 --subject_num 0 --fold_num 0
```

2. Training all subjects:
```bash
python ADFCNN/train_all_subjects.py
```

3. Evaluation:
```bash
python ADFCNN/evaluation.py
```

## Dataset
The BCI Competition IV 2a dataset contains EEG recordings from 9 subjects performing four different motor imagery tasks:
- Left hand movement
- Right hand movement
- Feet movement
- Tongue movement

Each subject participated in two sessions: training and evaluation.

## Citation
If you use this code in your research, please cite:
```
[Citation information will be added]
```

## License
This project is licensed under the MIT License - see the LICENSE file for details.

## Acknowledgments
- BCI Competition IV 2a dataset providers
- Contributors and researchers in the BCI community

ADFCNN-MI

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README