About
I am a machine learning engineer at Omniscience.
Before I started working at Omniscience, I was doing computational neuroscience researches, interested in the intersection of these areas. I have mainly worked with EEG, but I am also interested in other areas. My interest is on the application of machine learning to neuroscience data.
Past Projects
Finger Movements Decoding with EMG
As computers diversify, there is an increasing need for different control interfaces. Electromyography (EMG) is a potential new interface to computers which is starting to be explored by commercial companies. EMG might be able to control computers without physically touching anything and without requiring space. However, there is no consensus on which algorithms best allow classification of movement from the EMG signal. I found that Random Forests using time domain EMG data was the best within-participant classifier (79% accuracy for 4 different finger movements). The results were consistent with the consensus of the field, but this is one of the first reports of finger movement classification using Random Forests. This result confirms the possibility of using EMG as an interface to computers but raises further challenges to improve the accuracy.
ScreenPortal
Mind Wandering Detection with EEG
Github link for the modified game