Special Lectures


By alphabetical order (last name) 

Machine Learning Approaches in Probing Neural Control of Skilled and Disordered Movements in Music Performance
 

Shinichi FURUYA, Ph.D

Post-doctoral Fellow, 
Institute for Music Physiology and Musicians' Medicine, 
Hannover University of Music and Drama, Germany
Abstract: 
  The human motor system has redundant number of degrees of freedom (joints and muscles), which enables a variety of movement production. Body movements change in relation to task demand, intention, learning, and development of disorder. To decode them through examining movements provide insights into neural and biomechanical principles governing skilled, unskilled, and disordered movements, an approach referred to as “reverse engineering”. We have tried to address this issue by studying the upper-extremity movements of healthy pianists, pianists with focal hand dystonia, and non-musicians using machine learning techniques. The presentation will focus on introducing effectiveness and usefulness of multivariate and cluster analyses for probing neural control of skilled and disordered motor behaviors. For example, principal component analysis and EM algorithm successfully identified a smaller number of hand movement patterns while pianists were playing variety of musical pieces, which simplified the control of the movements of multiple degrees of freedom at the hand. Linear regression analysis and K-means clustering also allowed for characterizing individual differences across players according to the movement strategy to manipulate several acoustic variables during piano keystrokes. This also determined muscles with a high risk of repetitive strain injuries for individual pianists, which would be of help for the prevention of such movement disorders. Supervised classification techniques such as naïve Bayes and support vector machine turned out to be capable of specifying a finger affected by focal dystonia based on spatial and temporal features of individuated finger movements, which implicates that these techniques are applicable for accurate clinical diagnosis. These findings suggest strong potentials of an inter-disciplinary approach linking machine learning and movement science for helping people who need outstanding motor skills.

Biography: 
  A post-doctoral fellow at Institute for music physiology and musicians’ medicine (IMMM), Hannover university of music, drama and media (HMTMH), Germany. At Osaka University, he earned a bachelor degree at Department of Engineering Science, master degree at Graduate School of Human Science, and PhD at Graduate School of Medicine. After working with 3D motion capture projects at Kwansei Gakuin University, he started to work at Department of Neuroscience in University of Minnesota, where he studied sensory-motor control of finger movements in piano playing. His research focus ranges from biomechanics and motor control of skilled, unskilled, and disordered movements in piano playing through development of diagnosis and neuro-rehabilitation techniques for focal hand dystonia, a neurological disorder that potentially terminates professional careers of musicians. His expertise includes dynamic analysis of multi-linked body movements by using inverse and forward dynamics techniques, multivariate analysis to probe neural mechanisms behind control of redundant number of joints and muscles, system development for real-time manipulation of auditory feedback in musical performance, and clinical intervention using brain stimulation. He received awards including a finalist of young investigator award of International Society of Biomechanics (ISB), scholarship winner of Society of Neural Control of Movement (NCM), JSPS Research Fellowship for Young Scientists, Research Fellowship for Postdoctoral Researchers by Alexander von Humboldt Foundation, and JSPS Postdoctoral Fellowship for Research Abroad. Currently, he works as a guest associate editor at Frontiers in Human Neuroscience and Frontiers in Auditory Cognitive Neuroscience. As a pianist, he won prizes at several professional piano competitions such as Kobe International Piano Competition and Japan Classical Music Competition. (website) www.neuropiano.net

 

The K Computer and AICS


Dr. Kimihiko Hirao

Director, 
Advanced Institute for Computational Science,
RIKEN, Japan
Abstract: 
  The K computer starts the official operations in September, 2012. The K computer won the top position on TOP500 last year achieving a LINPACK benchmark performance of 10 petaflops - becoming the first supercomputer ever to reach this milestone. Computer simulation is becoming more and more important for contemporary science and engineering. Simulations performed on the supercomputer will drive progress in science and technology and play an important role in solving difficult problems that we face as a society. There are very critical issues that need to be solved - global warming, alternative energy, disaster mitigation, healthcare, security, etc. The role of simulations will become increasingly larger, and the results that they provide will undoubtedly greatly affect society.

  Our institute, Advanced Institute for Computational Science (AICS) was established in July 2010. It is an organization charged with operating and managing the K computer and pursuing research and development by promoting strong collaborations between computer scientists and computational scientists. We have now 18 research teams in computer science and computational science. Plotting and developing Japan's strategy for HPC, including defining the path to Exascale computing, is also the important mission of AICS.

  On March 11th, 2011, East Japan was shaken by a massive earthquake. Homes were destroyed and whole villages swept away by the tsunami. The task of reconstruction has been made even more difficult by the Fukushima nuclear crisis. These crises have forced scientists to confront very tough issues. In the pursuit of ever-better lifestyles, we have not always made the best choices. We need to decide what kind of society we want to create, and this means defining new paradigms. We Japanese have stayed strong and worked together to overcome these crises. We would like to make the K computer a catalyst for the restoration of Japan. The true value of the K computer will be put to the test in the years ahead.

Biography:
  Born on November 9, 1945 in Niihama, Japan, obtained B.S. in 1969 and Ph.D in 1974 both from Kyoto University. A postdoctoral at University of Alberta (Prof. Sigeru Huzinaga) in 1974-1975 and worked with Prof. Roy McWeeny at Sheffield University in 1978-1979. Professor at Nagoya University in 1988 and moved to University of Tokyo in 1993. A Visiting Professor at Institute for Molecular Science (IMS) in 1989-1990 and 2002-2003. A Visiting Professor of University of the Air from 2003-2007. Dean of School of Engineering (2004-2006), Vice-President (2007-2009), Professor Emeritus, University of Tokyo (2009 - ), Special Adviser of Riken (2009-2010), Director of Advanced Institute for Computational Science, RIKEN (2010- ) Author of more than 320 scientific articles in Physical & Theoretical Chemistry.

 

A Fully Automated Evolutionary Art
   

Prof. Tatsuo Unemi

Department of Information Systems Science,
Soka University, Japan 
Abstract: 
  Owing to recent innovations in computer hardware and software, it has become possible to implement automated evolution to produce interesting abstract images based on computational aesthetic measures for fitness criteria within a feasible time using a personal computer. The lecture introduces the author’s recent projects on a fully automated evolutionary art. The recent progress of graphics processing unit (GPU) realized not only smooth drawing of 3D animation on the full hi-definition TV screen but also hi-speed image processing in real-time. The genotype used by the author’s software, SBArt4, is in a form of mathematical expression that defines a function to map the pixel’s spatiotemporal coordinate into the position in a color space. By compiling the genotype into a code fragment in a shader language, the generation of abstract image and the abstract image effects became possible to be completed within a short time for a standard frame rate on a hi-resolution screen. It also made the image processing fast enough to evaluate an image by a combination of a number of different types of statistical analysis to measure aesthetic values within a feasible time.

  Using two small computers connected via Ethernet, we realized an installation of automatic video art that continuously displays a series of new interesting animations on the screen by selecting individuals of higher fitness from an evolving population in real time. The art project also includes automated creation of animations that are accessible over the Internet. A personal computer at the author’s laboratory has added ten 20-seconds productions on the website everyday since October in last year, in a form of not only a movie file but also a shading code for WebGL that realizes the viewer’s experience of hi-resolution animation without any loss by image compression.

Biography: 
  He was born in Kanazawa, Japan in 1956. He graduated from Department of Control Engineering, Tokyo Institute of Technology in 1978, received Master's degree from Department of System Sciences in 1980, and Doctor's degree in 1994 from the same university. He worked as a research associate from 1981 to 1987 at Tokyo Institute of Technology, as an assistant professor at Nagaoka University of Technology, as an assistant professor at Soka University from 1992, as an associate professor from 1995, and then the current position from 2012. He was also working at Laboratory for International Fuzzy Engineering Research from 1992 to 1995 as a visiting scholar leading the Robot project. He was staying at AI laboratory in University of Zurich as a visiting professor from April to September 2000. His research included Natural Language Processing, Knowledge Engineering, Machine Learning, Genetic Algorithm, Reinforcement Learning, Distributed Autonomous Robot System, and Artificial Life. Current interests include artistic, sociological, and humanities applications of these technologies.

  Since 2003, he has worked with Swiss-based artist, Daniel Bisig, for new-media art projects. Their works have been exhibited and demonstrated in international events related to new-media arts, such as ARCO 2007 in Madrid, 10th Japan Media Art Festival in Tokyo, Art Escapes 2007 in Valencia, SIGGRAPH 2009 in New Orleans, ISEA 2010 in Dortmund and WRO 2011 in Wrocław. They received three awards, Honorary Mention in Vida 9.0 in 2006 by Flocking Messengers, Excellence Award in 10th Japan Media Art Festival in 2006 by MediaFlies, and Audience Prize in WRO 2011 by Cycles. They were involved in four projects of contemporary ballet by Jiří Kylián in 2008 and 2009 for the stage effects utilizing computer-based projection in The Hague and Munich.