[Project] EEG/Biosignal Classifier App

[mori-c]

Idea

Brain-Computer Interface classification feature could identify lobe activity strengths and channel output averages per node using FFT algorithm.

If time permits within study group schedule, add sentimental analysis as a soft reinforcement learning output

Goal

Create an EEG Biosignal classifier with PyTorch/Tensorflow connected with Firebase

TODO

• [ ] find out how to connect pytorch/tensorflow models into ios

Process

• jupyter notebook (code)
• firestone, firebase (database)
• pytorch (library simulator)
• flask (server requests)
• django (frontend)
• muse api (ios)

Backup:

in case pytorch models don't integrate well

• colab notebook (code)
• firestone, firebase (database, server requests)
• tensorflow (library simulator)
• angularJS (frontend)
• muse api (ios)

Breakdown

✅	Process	I / O	Method / Tools
	store/upload data	input	firestone
	clean eeg data	input	colab, numpy, pandas
	graph, visualize data	output	colab, matplotlib
	identify frequency channels	...	...
	remove interfernce noise	input	firestone, csv
	apply deep learning algorithms (FFT)	input	colab, algebra
	display output of either image or wavelengths (tbd)	output	...
	identify loss perhaps around less active channels by boosting (reinforce activity with auditory engagement, for example)	...	colab, generalization, loss, monte carlo / bayesian / regression
	> to be continue	...	...

Firestone (database) → Firebase Setup

Firestore Sub-Collections

Framework

Django or Angular

Repo

Documentation and code found under mori-c/meetups/gdg

Notes

Based on neurotechnology coursework completed recently will inspire most of the methods used to execute this webapp