Build an attractive new preprocessing and classification method for public dataset

We are working on a project on developing preprocessing and classification method of hybrid EEG-NIRS data from this link https://doc.ml.tu-berlin.de/simultaneous_EEG_NIRS/ and https://doc.ml.tu-berlin.de/hBCI/. Take a look at the publications in the attach files. The original code can be accessed from https://github.com/JaeyoungShin/simultaneous_EEG-NIRS. We would like to develop a new preprocessing and classification algorithm using the machine or deep learning model with any language, then apply it to the 2 datasets (run each sample separately for both datasets). Further, design an attractive architecture for the entire experiment. The result format will be very similar to the attached paper. The classification accuracy should be higher than the attached papers. Please contact me for more detail.