Duke Enable

I led a research and development team in Duke eNable, where we aim to develop an algorithm for predicting intended hand postures from muscle signals alone. We started by learning about basic concepts machine learning, such as back-propagation, convolution, and transfer learning to prepare team members for developing the algorithm. We then went through relevant literature about existing algorithms so that we can modify and improve them for better performances. Currently, the algorithms are being developed on open source electromyographic (EMG) datasets that were collected using the Myo armband. We hope to test these algorithms on EMG signals from our team members or amputees in the near future.
Relation to Grand Challenge:
This experience gave me the opportunity to develop a product for decoding the peripheral signals of the nervous system that can be used for more convenient prosthetics for amputees. Through this experience, I learned to expand on an existing problem, acquire the relevant knowledge or information, and to organize a team to create a solution and implement it.
Supervisor: Michael Faber
Start Date: 8/1/2018
End Date: 3/20/2020
Hours: 180