This document discusses 11 applications of machine learning to music research, focusing on expressive music performance. It describes two approaches - learning at the note level and learning at the structure level. For the note level approach, it uses a system called IBL-SMART that learns rules to determine loudness and tempo for each note. For the structure level approach, it analyzes musical structures like phrases and learns prototypical expression shapes associated with them. It presents experiments applying these approaches to classical pieces, finding the structure level approach produced more musically convincing results.