Beck, Joseph E.
This paper examines how the changes in factors of a deep learning network can affect the output. The experiments were undertaken by modifying the features of deep learning network in order to view the differences in output. It was found that the single output network has a better performance on most of the output values, while multiple output network has a better root-mean-squared error performance and an obvious decrease in wheel spin.
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.