Joe Cleveland '20 and his research advisor Prof. Doug Turnbull (CS) presented their work on using triplet neural networks to model acoustic music similarity at the Machine Learning for Media Discovery Workshop (ML4MD) as part of the International Conference on Machine Learning (ICML) on July 18th, 2020.
Joe's work was selected for a full 15-minute oral presentation at a workshop. This honor was given to just seven of the fifteen papers that were accepted for publication. The workshop was attended primarily by graduate students, postdocs, faculty members, and industry researcher from companies such as Spotify, Pandora, Apple, and Netflix.
This work was a collaboration with Derek Cheng '20, Michael Zhou '22, and Prof. Throsten Joachims at Cornell University. It is related to some of the work that Joe did last summer as an H&S Summer Scholar. This ongoing work is currently supported by NSF grants IIS-1615706/1615679 and IIS-1901168/1901330.
The research will be used to improve playlist generation algorithms for the Localify.org project at Ithaca College. Localify.org is a locally-focused music event recommendation platform. It was taken down due to Covid-19 (i.e., no live shows to recommend) but will be relaunch this fall with a renewed focus on promoting local music scenes.
Paper details:
Content-based Music Similarity with Triplet Networks
Joe Cleveland, Derek Cheng, Michael Zhou, Thorsten Joachims, Doug Turnbull
International Conference on Machine Learning (ICML)
Machine Learning for Music Discovery Workshop (ML4MD)
July 2020
https://www.ithaca.edu/intercom/article.php/20200805100627767