Week 21 (04/04/2021)

For this week, the team attempted to figure out the issue with the multiplexer. To test it out, we connected two power supplies to the multiplexer channels to test it out. The oscilloscope did not illustrate the same issue from the previous week. We connected a unity gain buffer between the output of the multiplexer and the input of the instrumentation amplifier, but it did not resolve the issue. We will continue testing the issue with the multiplexer by connecting a function generator with a small amplitude as one of the inputs for the multiplexer, while another input will be one of the electrodes connected to one of our heads.

Besides debugging the circuit, the team focused on data collection. Approximately 30 "yes" questions and 30 "no" questions were used to collect Tyler and Mark's data. The questions were repeated to have two "yes" datasets and two "no" datasets per person. In total four datasets per person was collected.

Figure 1 illustrates the k-means clustering that was performed on the collected data. The black markers (circles) in Figure 1 are the centroid locations for the 'yes' datasets, while the lime green markers (circles) are the centroid locations for the 'no' datasets. Centroids are points at which a majority of the points are centered at. As more datasets are collected, the newly collected datasets will be compared with these points to determine whether the centroids accurately depict 'yes' or 'no'. The machine learning algorithm will be adjusted accordingly with the newly collected data as well to improve centroid locations.

Figure 1: K-mean clustering of "yes" and "no" datasets

THIS WEEK

Tyler

• Collected further data using the EEG Helmet for further analysis

• Performed testing and debugging on the multiplexers to determine the issue with the multiplexer switching

Mark

• Collected further data using the EEG Helmet for further analysis

• Performed k-means clustering on collected Yes and No datasets

NEXT WEEK

Tyler

• Continue to debug the multiplexer

• Potentially assemble the EEG helmet

Mark

• Collect further data using the EEG helmet for further analysis

• Perform different methods of k-means clustering on the data collected for comparison against previously collected data

Memo: Dr. Hossein Asghari further reviewed the multiplexer and assisted the team in determining what the issue may be. After further discussion, the team has considered that the spikes we have been seeing after the multiplexer switches are the settling time.