0:00/0:00
April 14, 2005 01 h 01 min
April 14, 2005 24 min
May 12, 2005 52 min
February 4, 2005 01 h 18 min
October 17, 2007 49 min
June 27, 2007 01 h 12 min
July 11, 2007 48 min
September 12, 2007 01 h 07 min
September 19, 2007 01 h 13 min
September 26, 2007 01 h 00 min
October 3, 2007 01 h 12 min
October 10, 2007 01 h 10 min
October 24, 2007 50 min
November 21, 2007 57 min
How can machine learning help you write a song? In this talk, I will present two projects that grew out of February Album Writing Month (FAWM.ORG), an international online community of musicians that I help organize. Each year, the goal is to compose 14 new pieces of music during the month of February.
First, I will describe a novel machine learning approach to modeling and and predicting how songwriting collaborations form and succeed within the community using random walks on the social network graph. Because collaborations are associated with positive user outcomes (number of songs written, reaching the 14-song goal, etc.), these results give us insight into how we can help members have a better FAWM experience.
Second, I will discuss two generative language models that we used to create a suite of "computational creativity tools" called The Muse. Despite their statistical simplicity, these tools have successfully helped hundreds of songwriters find and develop ideas for new songs.
I will conclude by proposing a few open directions for how machine learning can support individual and group creativity.
[Part of this research is joint work with Steven Dow.]
Burr SETTLES, invité par l'équipe du projet MuTant (c/o Représentations Musicales) présente :
"Machine Learning and Computer-Assisted Creativity"
Abstract:
How can machine learning help you write a song? In this talk, I will present two projects that grew out of February Album Writing Month (FAWM.ORG), an international online community of musicians that I help organize. Each year, the goal is to compose 14 new pieces of music during the month of February.
First, I will describe a novel machine learning approach to modeling and and predicting how songwriting collaborations form and succeed within the community using random walks on the social network graph. Because collaborations are associated with positive user outcomes (number of songs written, reaching the 14-song goal, etc.), these results give us insight into how we can help members have a better FAWM experience.
Second, I will discuss two generative language models that we used to create a suite of "computational creativity tools" called The Muse. Despite their statistical simplicity, these tools have successfully helped hundreds of songwriters find and develop ideas for new songs.
I will conclude by proposing a few open directions for how machine learning can support individual and group creativity.
[Part of this research is joint work with Steven Dow.]
Bio:
Burr Settles is a Data Scientist and Software Engineer at Duolingo, a crowdsourcing ecosystem that combines foreign language education and translation. He also runs the website FAWM.ORG, an annual songwriting challenge for musicians worldwide. Previously, he was a postdoc in Carnegie Mellon's Machine Learning Department, and earned a PhD in Computer Sciences from the University of Wisconsin-Madison. His research focuses on interactive machine learning that resembles a "dialogue" between computers and humans, with applications in natural language processing, creativity, and social computing. He recently organized workshops at the ICML and NAACL conferences on such learning strategies. His book Active Learning (a short introduction to the field) was published in 2012 by Morgan & Claypool. In his spare time, Burr also plays guitar in the Pittsburgh pop band Delicious Pastries.