Motor Imagery for two-class task classification

We will use motor imagery data for training a machine learning model capable of discerning between two classes. This prediction can be used to potentially move objects with our minds.

A 30.000ft view of the process

1. Import data
2. Preprocessing (load -> highpass, lowpass, notch filter -> (csp) -> artifacts -> psd -> make epochs)
3. Feature extraction (load epochs -> get mu band -> average mu band -> make feature)
4. Model training & prediction (train/test split -> training -> prediction)

Data IO (data shapes and formats) for each stage

1. Import data

• Input: MOABB
• Output: .fif -> ndarray[15, 160*5*512] -> [channels, readings]

2. Preprocessing (load -> highpass, lowpass, notch filter -> (csp) -> artifacts -> psd -> make epochs)

• Input: .fif -> ndarray[15, 160*5*512] -> [channels, readings]
• Output: ndarray[15, 512*5, 160] -> [channels, epoch_length, n_epochs]

3. Feature extraction (load epochs -> get mu band -> average mu band -> make feature)

• Input: ndarray[15, 512*5, 160] -> [channels, epoch_length, n_epochs]
• Output: ndarray[15, 160] -> [channels, epoch_channel_mu_band_average]

4. Model training & prediction (train/test split -> training -> prediction)

• Input: ndarray[15, 160] -> [channels, epoch_channel_mu_band_average]
• Output: prob. class