TY - CONF
T1 - Towards deep interpretability (MuS-rover II)
T2 - 5th International Conference on Learning Representations, ICLR 2017
AU - Yu, Haizi
AU - Varshney, Lav R.
N1 - Funding Information:
We thank Professor Heinrich Taube, President of Illiac Software, Inc., for providing Harmonia’s MusicXML corpus of Bach’s chorales (https://harmonia.illiacsoftware.com/), as well as his helpful comments and suggestions. This work was supported by the IBM-Illinois Center for Cognitive Computing Systems Research (C3SR), a research collaboration as part of the IBM Cognitive Horizons Network.
Funding Information:
We thank Professor Heinrich Taube, President of Illiac Software, Inc., for providing Harmonia's MusicXML corpus of Bach's chorales (https://harmonia.illiacsoftware.com/), as well as his helpful comments and suggestions. This work was supported by the IBM-Illinois Center for Cognitive Computing Systems Research (C3SR), a research collaboration as part of the IBM Cognitive Horizons Network.
Publisher Copyright:
© ICLR 2019 - Conference Track Proceedings. All rights reserved.
PY - 2017
Y1 - 2017
N2 - Music theory studies the regularity of patterns in music to capture concepts underlying music styles and composers' decisions. This paper continues the study of building automatic theorists (rovers) to learn and represent music concepts that lead to human interpretable knowledge and further lead to materials for educating people. Our previous work took a first step in algorithmic concept learning of tonal music, studying high-level representations (concepts) of symbolic music (scores) and extracting interpretable rules for composition. This paper further studies the representation hierarchy through the learning process, and supports adaptive 2D memory selection in the resulting language model. This leads to a deeper-level interpretability that expands from individual rules to a dynamic system of rules, making the entire rule learning process more cognitive. The outcome is a new rover, MUS-ROVER II, trained on Bach's chorales, which outputs customizable syllabi for learning compositional rules. We demonstrate comparable results to our music pedagogy, while also presenting the differences and variations. In addition, we point out the rover's potential usages in style recognition and synthesis, as well as applications beyond music.
AB - Music theory studies the regularity of patterns in music to capture concepts underlying music styles and composers' decisions. This paper continues the study of building automatic theorists (rovers) to learn and represent music concepts that lead to human interpretable knowledge and further lead to materials for educating people. Our previous work took a first step in algorithmic concept learning of tonal music, studying high-level representations (concepts) of symbolic music (scores) and extracting interpretable rules for composition. This paper further studies the representation hierarchy through the learning process, and supports adaptive 2D memory selection in the resulting language model. This leads to a deeper-level interpretability that expands from individual rules to a dynamic system of rules, making the entire rule learning process more cognitive. The outcome is a new rover, MUS-ROVER II, trained on Bach's chorales, which outputs customizable syllabi for learning compositional rules. We demonstrate comparable results to our music pedagogy, while also presenting the differences and variations. In addition, we point out the rover's potential usages in style recognition and synthesis, as well as applications beyond music.
UR - http://www.scopus.com/inward/record.url?scp=85088228432&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85088228432&partnerID=8YFLogxK
M3 - Paper
AN - SCOPUS:85088228432
Y2 - 24 April 2017 through 26 April 2017
ER -