It's me on the photo's right wearing leetcode shirt collecting data and setting up the conditions for the Motor imagery experiment in a Local LAB
This dataset was collected by me as a means for training machine / deep learning models in an EEG motor imagery classification. This was one of my roles as a machine learning engineer for the graduation project in Faculty of Artificial Intelligence, KafrElsheikh university.
I am profoundly grateful to all the technical support and advice provided by the project's supervisor, Dr. Mona AlNaggar
The goal is to perform motor imagery classification (left, right, relaxed) and translate JUST thoughts into action.
I used Muse2 Headband with 4 electrodes and Muse Monitor android app to start the recording sessions, and export the CSVs.
This high dimensional, temporal-data was collected in both subject-independent and subject-dependent contexts with the help of 19 healthy subjects (12 males, 7 females) in different states aged between 19 and 68 as a means of training for various deterministic and non-deterministic machine learning models to carry out Motor imagery classification task. 20 columns, where we have 5 powerbands (Alpha, beta, theta, delta, gamma) per each of the 4 sensor-electrodes, were of significance to the motor imagery classification. I didn't use the raw data. However, raw data is exported via muse monitor too so you can use it as more insights can be extracted out of the raw data and thus use 4 columns only (AF7, AF8, TP9, TP10).
Features like gyro, accelerometer weren't an area of interest for this EEG brain analysis or motor imagery classification. Feature engineering techniques like PCA, ICA can be beneficial especially for the raw data scenario.
Motor Imagery is one class of the event-related potentials. It is the imagination of motion, without doing any actual movement.
| اسم المستقل | Muhammad E. |
| عدد الإعجابات | 1 |
| عدد المشاهدات | 13 |
| تاريخ الإضافة | |
| تاريخ الإنجاز |
