README.md

Evidence-of-Task-Independent-Person-Specific-Signatures-in-EEG-Using-Subspace-Techniques

Implementation of architecures proposed in https://ieeexplore.ieee.org/document/9383011

Major Requirements

• Matlab
  • Signal Procresing Toolbox
• Python (v3.6)
  • Keras (v2.2.2)
  • tensorflow (v1.9.0)

Instruction to train the models dicussed in paper

1. Install all major requirement and python packages given in requirements.txt
2. Clone the respository
3. Request the for data and download the same from here
4. Edit the configuration file to reflect the location of downloaded data
   • The configuration file can also be used to change classifiers (modified-i-vector (WIP), modified-x-vector and ix-vector (WIP))
5. Run the Run.m matlab script. This script with do the following
   • Split the diven data into train, validation and text
   • Extract features as per the configuration
   • Train the classifier as per the configuration
   • Report the obtained results on validation and test data

Note: The repository also includes a docker file which can be used to build a docker image with all requirements pre-installed.