This project investigates machine learning algorithms in the field of prosthetics. The machine learning model was created to predict finger movements when provided with an ultrasound image of the forearm. Convolutional neural networks, recurrent neural networks, support vector machines and k- nearest neighbor algorithms were developed and tested against one another. Upon completion of research, a convolutional neural network was deemed to be the most effective in this application. The code for the final convolutional network was submitted as an appendix (B) in this paper.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.