2016 Music and the Brain Symposium:
Music Information Retrieval and Data Science

Saturday, May 14 2016 | 10am–6pm
CCRMA (The Knoll), Stanford University [directions] [map]
This event is FREE and open to the public

Sponsored by the Scott and Annette Turow Fund

Click here to RSVP

Information for poster presenters

Partner event: Music and the Brain 2016—Resonance [info]
Sunday, May 15 | 10am–2pm | Li Ka Shing Learning and Knowledge Center, Stanford University
Co-sponsored by Stanford Music and the Brain and Stanford Music and Medicine


10:20amDouglas Eck, Google
Deep learning on large music datasets
11:10amTrevor Hastie, Stanford Department of Statistics
Sparse linear models
NoonLunch (on your own)
1:15pmKarthik Ramasamy, Twitter
2:05pmGert Lanckriet, University of California, San Diego
4:15pmElaine Chew, Queen Mary University of London
Music, mathematics, and models of the ineffable
5:05pmYe Wang, National University of Singapore
Neuroscience-informed sound, music, and wearable computing for rehabilitation and learning


Program organizers: Jonathan Berger, Blair Kaneshiro, Zhengshan Shi, Malcolm Slaney
CCRMA organizers: Nette Worthey, Matt Wright

Abstracts and Speaker Bios

Music, mathematics, and models of the ineffable
Elaine Chew, Queen Mary University of London

What makes music, really good music, difficult to model? Can structures that shape our understanding of music be gleaned from sonic information? Can music knowledge be formalized and represented mathematically? We shall consider these questions in the context of performance, the primary means by which music is communicated to the listener. An important function of performance is the use of prosodic variation to enunciate structure and to express musical form. The prosodic variations serve to segment music into coherent chunks or streams, to highlight pivotal moments, and to mark events of significance. Multiple plausible structural solutions almost always exist; the structure experienced is subject to the vicissitudes of individual perception, that of the performer and that of the listener. The performance thus shapes the listener's experience of the music, and in particular the listener's perception of the music's structure and meaning. What then are the hallmarks of a successful performance? While many have likened music composition to mathematical proof because of its pursuit of "unexpectedness, combined with inevitability and economy" (Hardy 1940), few have openly acknowledged the premium that performance places on these same qualities. We shall consider how music structures are found and chosen, and how successful designations of structural solutions—from articulations that accentuate motivic groupings to tipping points that signal the return from the edge of a musical excursion—manifest the properties of surprise, inevitability, and parsimony.

Elaine Chew is Professor of Digital Media at Queen Mary University of London, where she is affiliated with the Centre for Digital Music in the School of Electronic Engineering and Computer Science and co-leads the center's research theme in Music Cognition, Creativity, and Expression. Her research centers on computational modeling of music cognition and music performance, with the goal of revealing the thinking behind music performance. She was previously an Associate Professor at the University of Southern California where she was affiliated with the Viterbi School of Engineering and Thornton School of Music, and founded the Music Computation and Cognition Laboratory. In the US, She was a recipient of NSF CAREER/PECASE awards for research on Performer-centered Approaches to Computer-assisted Music Making; she was also the 2007-2008 Edward, Frances, and Shirley B. Daniels Fellow and leader of a research cluster on Analytical Listening through Interactive Visualization at the Radcliffe Institute for Advanced Study at Harvard University. She is author of a 2014 Springer monograph on Mathematical and Computational Modeling of Tonality: Theory and Applications. Elaine Chew received PhD and SM degrees in Operations Research at MIT, a BAS degree in Music Performance (distinction) and Mathematical and Computational Sciences (honors) at Stanford University, and Fellowship and Licentiate diplomas in piano performance from Trinity College London. She performs as soloist and chamber musician, and has worked with composers to create, premiere, and record new compositions.

Deep learning on large music datasets
Douglas Eck, Google

I'll talk about the Magenta Project, an effort in Google Brain to generate music, images, video and text using machine learning. Our research goal is to better understand how generative processes can inform our understanding of media creation. This work relies heavily on ingesting and learning from large music data sets. I'll try to summarize the state-of-the-art in relevant generative models and focus on how this work relates to music information retrieval and data science.

Douglas Eck is a Senior Staff Research Scientist at Google, working on the Magenta project, an effort to generate music, video, images and text using machine intelligence. Magenta is part of the Google Brain team and is using TensorFlow (www.tensorflow.org), an open-source library for machine learning. The question Magenta asks is, "Can machine intelligence help musicians and artists be more creative? If so, how? If not, why not?" The goal of Magenta is to produce open-source tools and models that help creative people be even more creative. Dr. Eck is primarily looking at how to use "generative" machine learning models to create engaging media. Additionally, he's working on how to bring other aspects of the creative process into play. For example, art and music is not just about generating new pieces. It's also about drawing one's attention, being surprising, telling an interesting story, knowing what's interesting in a scene, and so on. Before joining Google in 2010, Douglas was an Associate Professor in Computer Science at the University of Montreal.

Sparse linear models
Trevor Hastie, Stanford Department of Statistics

In a statistical world faced with an explosion of data, regularization has become an important ingredient. In many problems, we have many more variables than observations, and the lasso penalty and its hybrids have become increasingly useful. This talk presents a general framework for fitting large scale regularization paths for a variety of problems. We describe the approach, and demonstrate it via examples using our R package GLMNET. We then outline a series of related problems using extensions of these ideas.

Trevor Hastie is the John A Overdeck Professor of Statistics at Stanford University. Hastie is known for his research in applied statistics, particularly in the fields of data mining, bioinformatics and machine learning. He has published five books and over 180 research articles in these areas. Prior to joining Stanford University in 1994, Hastie worked at AT&T Bell Laboratories for 9 years, where he helped develop the statistical modeling environment popular in the R computing system. He received his B.S. in statistics from Rhodes University in 1976, his M.S. from the University of Cape Town in 1979, and his Ph.D from Stanford in 1984.

Karthik Ramasamy, Twitter

Twitter generates billions and billions of events per day. Analyzing these events in real time presents a massive challenge. Twitter designed and deployed a new streaming system called Heron. Heron has been in production nearly 2 years and is widely used by several teams for diverse use cases. In this talk, we will give an overview of Heron, share our operating experiences and challenges of running Heron at scale.

Karthik Ramasamy is the engineering manager and technical lead for Real Time Analytics at Twitter. He is the co-creator of Heron and has more than two decades of experience working in parallel databases, big data infrastructure and networking. He cofounded Locomatix, a company that specializes in real time streaming processing on Hadoop and Cassandra using SQL that was acquired by Twitter. Before Locomatix, he had a brief stint with Greenplum where he worked on parallel query scheduling. Greenplum was eventually acquired by EMC for more than $300M. Prior to Greenplum, Karthik was at Juniper Networks where he designed and delivered platforms, protocols, databases and high availability solutions for network routers that are widely deployed in the Internet. Before joining Juniper at University of Wisconsin, he worked extensively in parallel database systems, query processing, scale out technologies, storage engine and online analytical systems. Several of these research were spun as a company later acquired by Teradata. He is the author of several publications, patents and one of the best selling book "Network Routing: Algorithms, Protocols and Architectures." He has a Ph.D. in Computer Science from UW Madison with a focus on databases.

Neuroscience-informed sound, music, and wearable computing for rehabilitation and learning
Ye Wang, National University of Singapore

The use of music as an aid in improving body and mind has received enormous attention over the last 20 years from a wide range of disciplines, including neuroscience, physical therapy, exercise science, and psychological medicine. We have attempted to transform insights gained from the scientific study of music, learning, and medicine into real-life applications that can be delivered widely, effectively, and accurately. We have been using music to enhance learning as well as to augment evidence-based medicine. In this talk, I will describe tools to facilitate the delivery of established music-enhanced therapies, harnessing the synergy of sound and music computing (SMC), mobile computing, and cloud computing technologies to promote learning and to facilitate disease prevention, diagnosis, and treatment in both developed countries and resource-poor developing countries. These tools are being developed as part of ongoing research projects that combine wearable sensors, smartphone apps, and cloud-based therapy delivery systems to facilitate music-enhanced learning and music-enhanced physical therapy. I will also discuss the joys and pains working in such a multidisciplinary environment.

Ye Wang is an Associate Professor in the Computer Science Department at the National University of Singapore (NUS) and NUS Graduate School for Integrative Sciences and Engineering (NGS). He established and directed the sound and music computing (SMC) Lab (www.smcnus.org). Before joining NUS he was a member of the technical staff at Nokia Research Center in Tampere, Finland for 9 years. His research interests include sound analysis and music information retrieval (MIR), mobile computing, and cloud computing, and their applications in music edutainment , e-Learning, and e-Health, as well as determining their effectiveness via subjective and objective evaluations. His most recent projects involve the design and evaluation of systems to support 1) therapeutic gait training using Rhythmic Auditory Stimulation (RAS), 2) second language learning, and 3) motivating exercise via music-based systems.

Information for poster presenters

Students from CCRMA, Stanford Statistics, and related departments are invited to present posters on research related to Music Information Retrieval and Data Science.

Exact time of poster session to be announced soon.

Poster dimensions: Not to exceed 44"H x 44"W
Presenters are welcome to use a layout of their choice or one of the following templates (.pptx):
[portrait] [landscape]
LaTeX users may wish to use Jorge Herrera's LaTeX poster template

Click here to sign up to present a poster

Questions about posters? Email Zhengshan Shi, kittyshi ~at~ ccrma \./ stanford \./ edu