What techniques have you had success with when it comes to signal processing?

I've taken up bio-hacking and working with BCI's as a hobby, but it is a completely new field to me. I have a background in programming, but no experience working with processing brainwaves. (I never thought I'd be saying that haha!) The OpenBCI GUI makes everything look easy, and I've managed to connect to my Cyton-daisy board using BrainFlow in Python. I have the ability to collect the data, and my question for the more experienced bio-hackers is what comes next? In the OpenBCI GUI the magic seems to happen behind the scenes, applying Notch Filters, Bandpass filters, and transforming the data into something that is cleaner. I've got these CSV files that contain just the raw data.

Sample Index, EXG Channel 0, EXG Channel 1, EXG Channel 2, EXG Channel 3, EXG Channel 4 ...
0.0, -5526.580575296948, -4722.856548173017, 3500.1937747232646, -724.5094447743231 ... 

The goal is to start playing with some machine learning algorithms, starting with simple things.. maybe predict whether my eyes are opened or closed, and move on to more complex things from there. I'm curious what other people have had success with? What techniques do you use to filter and de-noise the data? I assume the data in it's most raw format wouldn't be likely to yield very good training and classification results. As someone with very little experience with signal processing, I'm hoping some of the veterans in the field might weigh in. I know I could just start throwing random filters and signal-processing techniques at the data, and eventually I'd likely find something that works, but I'm thinking there are likely experts here that know the industry-standard techniques. What are other people doing to process the raw data to clean it up and prepare it for training and testing ML models?

Sign In or Register to comment.