Ryan Strelow

Recurrent Neural Network Design in Music Composition

Jeremy Iverson, Advisor
Imad Rahal
Peter Ohmann 

What inspired Ryan to select his thesis topic?
Creating aesthetically pleasing music by algorithmic composition has continually been an ambitious goal of music information retrieval research.  As a computer scientist, my interest lies in understanding the possible overlaps between computer science concepts and traditionally creative settings such as music. 

Ryan’s research in his own words:
This research presents different neural network designs for creating and classifying music.  Shown is the difference observed in performance and accuracy arising from how musical notes, and furthermore songs, are represented as data for neural networks.