NeuroTechSC

8/31 Week Updates

Hardware Team

In this week's meeting, the hardware team brainstormed solutions to many design issues. Some main concerns were how to deal with the weight of the batteries; placement of the board; and accommodating different head shapes. The latter is the most important since the placement of the electrodes is essential. It was suggested that the headset utilize a form of flexible tubing that retains its form after adjustment to house the electrodes. This topic is still up to debate and will be discussed with a graduate student. Furthermore, having a printed prototype by Wednesday was discussed. Having a physical headset will allow them to identify design flaws. Finally, a future goal is to have members of different teams make recordings after the headset is finalized to aid the data team.

Data Team

The data team has split into two groups, one focused on building the pipeline, and the other focused on processing the data. This week, they successfully set up the pipeline to read EMG data. Using the BrainFlow API, they wrote code that is able to read the data through python and is able to print it in the console. For now, they will use this for training purposes and it will allow for the processing group to start on writing their code to process the data for input into the ML model.

Machine Learning Team

Machine Learning has discussed a few different plans for the upcoming week. They have decided to split into a few teams of 2 to better focus on implementing CNN and RNN models. This will allow them to specialize more heavily in what they are familiar with, making the process easier for everyone involved. They are also implementing a workshop to get everyone up to speed with the concepts required to effectively collaborate in their team. On another note, they’ve been focusing on choosing between 1 or 2-dimensional convolutional data. This is important because 2-dimensional data shows a lot more, but is sometimes less accurate, hard to implement, and less efficient. They are progressing towards a working EEG model and should have it done within the end of the week hopefully.

UI Team

Our UI team is building a full-stack web interface, using the ReactJS framework for the frontend and Flask for the backend. The UI will display the question asked and the answer (Yes/No) from the hardware. Currently, they have some basic boilerplate which displays a random question and answer as a placeholder as the team familiarizes themselves with the tech stack, allocates tasks, and works on the implementation. The exact implementation details of the display will depend on the interface provided by the data team.