Some results on data transformation in machine learning and data mining

Speaker: Tu Bao Ho
Japan Advanced Institute of Science and Technology.

Abstract: Transformation of data originally collected and represented in one space to another space can allow us to better exploit the data. Traditional examples include Fourier transformation, wavelet transformation, principal component analysis (PCA), latent semantic analysis (LSA), etc. Data transformation becomes more crucial when we have to deal with big data. This talk presents our contribution to the research on data transformation in the recent areas of machine learning and data mining: kernel methods, probabilistic graphical models and non-negative matrix factorization.

Biography: Ho Tu Bao is a Professor, Head of Machine Learning and Data Mining Lab at Japan Advanced Institute of Science and Technology (JAIST). He received degrees of master (1984), PhD (1987) from Paris 6 University, and Habilitation à diriger des recherches (1998) from Paris 9 University, France. His research interests are mainly in machine learning, data mining, computational science and biomedicine informatics. He is author and co-author of numerous papers published in top-ranked journals and conferences.


Presentation slide: download

Computational instrumentation and imaging for industrial and medical applications

Speaker: Masayuki Fukuzawa
Department of Information Science, Graduate School of Science and Technology,Kyoto Institute of Technology, Matsugasaki, Sakyo-ku, Kyoto 606-8585, Japan

Abstract: Instrumentation and imaging are fundamental techniques for industrial and medical applications. Recent progress of computation enables us to extend their functionalities, to improve their performances, and to reduce their costs. This talk provides such recent and ongoing studies mainly in industrial visual instrumentation and medical imaging as follows: 1) Quantitative imaging of residual strain in commercial semiconductor wafers for development of their manufacturing process as well as quality control of them, 2) High-speed distance measurement by image-based techniques with active illumination for track vehicles and conveyance robots, 3) 3D visualization of pulsatile tissues in ultrasonic movies of neonatal cranium for assisting bedside diagnosis of ischemic diseases, and 4) Computer-aided 3D-shape construction of hearts from CT data for surgical simulation. These theoretical backgrounds and experimental results are presented to demonstrate the potential of our techniques and the future prospect of such computational approaches.


  1. M. Fukuzawa, R. Kashiwagi, M. Yamada, Physica, Status Solidi (c), C 8, No. 2, 432-434, 2011
  2. K. Jiptner, M. Fukuzawa, Y. Miyamura, H. Harada, K. Kakimoto, and T. Sekiguchi, Solid State Phenomena, Vols. 205-206, pp. 94-99, 2013.
  3. M. Fukuzawa, H. Hama, N. Nakamori, and M. Yamada, Proc. of 5th Intr. Conf. on Electrical and Computer Engineering (ICECE2008), 412-416, IEEE Catalog #CFP0868A-CDR (2008).
  4. M. Fukuzawa, H. Hama, N. Nakamori, and M. Yamada, Proc. of 11th Intr. Conf. on Computer and Information Technology (ICCIT2008), 516-520, IEEE Catalog #CFP0817D-CDR (2008).
  5. M. Fukuzawa, M. Yamada, N. Nakamori, and Y. Kitsunezuka, Proc. of SPIE 6920, 69200S-1-8, 2008.
  6. M. Fukuzawa, K. Kawata, N. Nakamori, Y. Kitsunezuka, Proc. SPIE Vol.7968, 79681A-1-7, 2011.
  7. I. Shiraishi, M. Yamagishi, K. Hamaoka, M. Fukuzawa, T. Yagihara, Eur J Cardiothorac Surg. 37, 302-306, 2010.
  8. M. Fukuzawa, Y. Kato, N. Nakamori, S. Ozawa, I. Shiraishi, Proc. SPIE Vol. 8290, 829017, Feb. 2012.

Agent-based simulation to study complex socio-environmental systems

Speaker: Patrick Taillandier
The international UMMISCO research unit (team MSI – Hanoi, Vietnam).

Abstract: In many domains (land-planning, epidemiology,...), being able to model and simulate a socio-environmental system is mandatory. One of the most promising approaches to study and analyze such systems is agent-based modeling (ABM). This approach consists in modeling the studied system as a collection of interacting decision-making entities called agents. An agent-based model can exhibit complex behavioral patterns and provide relevant information about the dynamics of the real-world system it represents. Moreover, it can be used as a virtual laboratory to test scenarios and support decisions making. A recent tendency concerning ABM is the development of more and more realist models that integrates large sets of data. This recent trend requires the ability to design and manage more descriptive and detailed models. Building complex, incremental, data-driven modular models with most of agent-based modeling platforms is a difficult task. The GAMA modeling and simulation platform, developed since 2007 as an open-source project, aims at answering this challenge by providing modelers - which are not, most of the time, computer scientists - with tools to develop and experiment highly complex models through a well-thought integration of agent-based programming, geographical data management, flexible visualization tools and multi-level representation. In addition, since its last version, it integrates a new graphical modeling editor allowing to ease the development by non-computer scientists and support participatory modeling.  GAMA is used in many research projects concerning different application (traffic and mobility, tsunami evacuation, dengue spreading, water resource management, urban expansion…).

Short biography:

Patrick Taillandier graduated in artificial intelligence from the University of Lyon 1 (France) in 2005 and received his PhD degree in 2008 at the University Paris Est. His PhD, which was carried out in the COGIT Laboratory of IGN-France (the French NMA), deals with the problem of automatic knowledge revision in systems based on trial and error methods and its application to cartographic generalization. Between 2009 and 2010, he worked in the international UMMISCO research unit (team MSI – Hanoi, Vietnam) on agent-based simulation, and more particularly on the development of tools to integrate GIS (Geographic Information System) data in the GAMA simulation platform. In 2011, he worked at the IRIT laboratory in Toulouse (University Toulouse 1 – France) on the MAELIA project that aims at using agent-based modeling to study the socio-environmental impacts of norms of management and governance of renewal natural resources. Since the end of 2011, he is professor assistant (with a CNRS-Higher Education chair) at the department of Geography of the University of Rouen (UMR IDEES), where he is still working on the use of agent-based modeling and simulation to study socio-environmental systems. Patrick Taillandier is the coordinator of several projects and networks dealing with agent-based modeling (ANR ACTEUR, PICS IMEA, RNSC SimTools Network, etc.).