I want to build a general pipeline for SSVEP EEG signal decoding, in which it is easy to complete basic EEG processing stages like:
- cutting and slicing your data
- filtering data
- applying feature extraction method
- matching pattern and get result
The key inspiration of the framework design ethic is modular, and I want to take the manually parts like experimental information, special filters and feature extraction methods out of the main executing. So that it's much easier to re-write or add new minds into the framework. In the other words, it is a framework that flexible, easy for new learner and try something new.
There are 3 main class defined in this framework. If you just focus on get model for online experiment or do some cross validation, the data_runner class and data_cross_validation class are what you need, you can just read through and run them. In the filter_apply class, you can configure your own time-filter parameters, and what I must admit is that the filter parameters in the current version is not the best, and haven't been optimized at all!
You may notice that there are several functions above the main classes, they can be named as helper functions. I can move them together and separate them to a helper collection and make the main program clear.