#24: Brian McFee: Music and Data Science

Dr. Brian McFee develops machine learning tools to analyze multimedia data. This includes recommender systems, image and audio analysis, similarity learning, cross-modal feature integration, and automatic annotation. As of Fall, 2014, he is a data science fellow at the Center for Data Science at New York University. Previously, he was a postdoctoral research scholar in the Center for Jazz Studies and LabROSA at Columbia University.

 

My conversation with Brian today was focused on discussing his research in music informatics and its many facets and applications. He tells about some of the methods he used during his dissertation, and I ask him for insight on how to get a recommender system to recommend stuff that you actually like.

 

Here are some of the highlights of the show:

 

[3:17] What came first for Brian, the data science or the music? [5:19] Of all the things he could have chose to study, why did Brian choose music? [7:35] What is it like to be in a branch of data science that has become so closely tied with industry and well understood by the public? [9:37] How has Brian’s work expanded his own taste in music, and given him an appreciation of jazz? [12:00] Brian gives a brief history of the field of music informatics. [14:48] Where was the field when Brian wrote his dissertation, “More like this: Machine learning approaches to music similarity?“ [17:15] How have the characteristics and features for making predictions about music evolved since then? [21:06] Why does the concept of genre generally irritate Brian, and what is the “David Bowie problem?” [26:20] How do you address the problem of subjectivity in the field when conducting research? [31:21] Is there a dilemma in trying to take a subjective art like music and trying to quantify it as a science? [35:24] How can a recommender system actually accurately predict what kind of music a listener is looking for? [38:43] What can you do to train your Spotify recommendations? [42:33] How do you make the career decision whether to stay in academia vs. go into industry? [46:31] What kind of problems is Brian currently interested in solving? [49:00] What major life lessons can be taken away from work in machine learning? [50:00] Rapid fire questions.

 

Enjoy the show!

 

Find more at www.ajgoldstein.com/podcast/ep24