Hardware Team

This week the hardware team has almost completed the new headset. This one is more narrow than the original. What needs to be completed next is the board and battery pack enclosing that will be mounted using a bicep strap. If all goes well this may be the final 3D print before the headset can be handed off to more team members to complete further recordings. According to how the headset fits the various members modifications may be needed.

• Cesar Ortega

Data Team

This week, the data team found that filtering, Independent Component Analysis (ICA), and standardizing the data yielded better results, as opposed to normalizing the data. Previously, the team was using normalization, a technique that changes the numeric values in a data set to have a common scale, similar to how you would normalize a vector. The standardization technique, on the other hand, rescales attributes to have a mean value of zero and a standard deviation of 1. Standardization is useful because it equalizes the range of values and data variability, eliminating the threat of a variable over-weighing another during analysis. The data team was able to reach an accuracy of 85 to 90 percent with these new techniques and changing with the variables. For now, they plan to add another data processing technique that can increase the accuracy even more. -Aracely Cano

Machine Learning Team

Overall, the club is trying to pivot to something more useful with unique applications. Something such as a device that allows people who can't speak to use the device and analyze their sub vocalizations to allow them to communicate. The goal for the ML team is for each person to come up with two different ideas for the applications of this device by Wednesday. The model is currently at 85-90% accuracy which is good, while filtering with ICS and standard scalar. The ML team goal to have a more generalized model where the accuracy is still high with more participants. For this week, onboarding with new members is being continued. The goal next week is to be training the model with new data. -Brandon Luu

UI Team

For the week of October 12, the UI team was able to fix their database issue they were struggling with. They have also been continuously working on connecting the hardware, which has been a challenge for the team. For the coming week, they will continue to work on this blocker, as well as finding a way to display the hardware port number on the interface. -Aracely Cano