It was a combination of the exciting hardware available, as well as a great deal of the interest in some of the disciplines available to use: particularly microservice architecture, web socketing for concurrent data transfer, and of course, machine learning! Furthermore, it seemed like a project which delved into some of the unanswered questions about how music affects us, and to see how much it really effects our mood and behavior.
What it does
Mu is a system built around a muse headband, which reads EEG brainwaves, and analyzes them with machine learning to categorize the brainwaves into a relaxed, focused, or excited state! From there, Mu will find songs which help you achieve that state more effectively over time, as you listen to more music, and allow it to analyze brain signals.
How we built it
Each member of our team has a different forte, and so we all delegated tasks to take advantage of that. We used a great deal of technologies we were already familiar with to set the stage for the newer things that we hadn't tried just yet, so we could spend more time really working and polishing those.
Challenges we ran into
This was a very challenging project, particularly with regards to managing how data was to be stored and handled, reading the different EEG inputs ( neuroscience is confusing), and, we ran into a fair bit of trouble implementing machine learning on the arcane dataset we received from the Muse Headband.
Accomplishments that we're proud of
I think we're all very proud of what we built, since its our first time doing machine learning, and applying it to the endeavor of mind reading.
What we learned
We continued to develop our skills with micro services and websockets, while learning some completely new things: namely things like redux, machine learning using scipy, and reading EEG brainwaves, which was really cool, and a lot of fun to work with.
What's next for Mu
Moving forward we'd like to make our data more dynamic, so that we can train our models using all of our users to deliver an even better experience. It would also be valuable data to analyze as an aggregate dataset, since it maps different songs to different brain wave frequencies, and it seems like an opportunity to learn a lot about the psychological effects of music.