TEAM 51 : Spykers | ntx-hack-23

Repository for Team 51's submission to NTX Hackathon 2023.
The demo video can be found here.

Objective: Testing if EEG can be used to deduce Focus

Getting Started

All code was tested on a Ubuntu 20.04 machine running Python 3.10. Required packages can be installed through:

pip install -r requirements.txt

Repository Navigation

read_serial.py

Reads serial EEG data from a configurable port and pickles the data for future use.

experiments/*.pkl

Recorded EEG signals used for hypothesis testing. These can be sub-divided as:

• record_*_focus.pkl: Baseline EEG activity.
• record_*_music_c.pkl: EEG activity when listening to music that the user likes.
• record_*_music_d.pkl: EEG activity when listening to music that the user does not like.
• record_sample*.pkl: EEG data of the user listening to various tracks.

experiments/compare*.py

Hypothesis testing under specific scenarios.

• compare_feature_usd.py: Tests to determine an indicator feature.
• compare_feature_likeness.py: Compares User Ratings and our Hypothesis for different tracks

music_player.py

Music player with auto-recommender system based off of our indicator feature. Still a work in progress.

results/*png

Hypothesis Testing results. These can be grouped as two types

• plt*.png: Hypothesis and Track Testing outputs produced by compare_feature_usd.py and compare_feature_likeness.py

• pb*.png: Correlation based indicator feature testing done using Backyard Brain's Spike Recorder.

Results

Hypothesis Tests

In our intial testing, we used this plot to determine the beta-alpha energy ratio as a decent indicator function:

To verify our hypothesis, we compared different tracks, which matched our hypothesis:

The results for correlation tests can be found under the results folder.