creating information utilization platform by integrating ... · creating information utilization...
TRANSCRIPT
Research supervisor:
Naonori Ueda
Creating information utilization platform by
integrating mathematical and information
sciences, and development to society
(Fellow, NTT Communication Science Laboratories;
Deputy Director, RIKEN Center for Advanced Intelligence Project)
2
1.BackgroundMathematical
Information
Platform
• The efficiency of using big data has been verified in various fields; recently, the AI
technology represented by deep learning (machine learning technology) is
attracting attention as a driver. However, in the case of rare events whose data are
difficult to collect, such as severe diseases, abnormal symptoms, and large-scale
disasters, or in the case where it is difficult to convert information into digital data,
we cannot say that the data driven-type approach alone is enough.
• In addition, the current technology of AI is called a black box model; building
explainable and trustworthy AI is also an important issue.
• On the other hand, in the field of science, the research based on a process model,
which involves elucidation of principles based on mathematics and/or mathematical
science, has been in practice since old days, where the ability of mathematics and
mathematical science has been verified.
• As for the field of information science, RSA cryptosystem, which was created in
Europe and the US; PageRank; compressed sensing; differential privacy; and other
innovative technologies can be said to be the results of the use of mathematics.
3
2.Overview
We aim to create a new platform technology from the cooperation and
integration of mathematical and information sciences with respect to the
solution of the problems in the actual society and the creation of added
values that may be difficult only with the data driven-type approach, such
as AI and big data analysis.
More specifically, we cope with the following research and development
areas.
1. Construction of the theory and technology that contribute to the creation of a
new information utilization method involving the concept of mathematics
2. Creation of a new science that links mathematics, mathematical science,
and information science
3. Creation of a next-generation application platform technology of data
analysis algorithms and software programs that accelerates and enhances
the utilization of information in various fields and the industrial world
We aim to lead to the solution of social problem with some impact, as mentioned
above.
Mathematical
Information
Platform
4
Information Science
Mathematical logic
Natural language processing
Dimensionality reduction
PageRank
Differential privacy
Software engineering
Computational science
Agent model
Network analysis
Deep learning
Security
Concealment calculationData mining
Machine learning
Graph theory
AlgorithmOptimization
Statistics
Data assimilation
Combinatorics
Topological data analysis
Compressed sensing
Geometry
Applied mathematics
Algebra
Mathematical engineering
Numerical analysis
Mathematics
Mathematical sciences
Analysis
LLL algorithm
RSA cryptosystem
3.Objectives and Research Area Overview
Medical MaterialBiologicalSocial &
Economics
Environmen
tal & Energy
Solving Social
Problems
Application
Fields
Mathematical Information
Platform
Science that connects
info. and math. sciences
New concept /
new theory
Based on MEXT material
• Creating information utilization platform by integrating mathematical and
information sciences
• Development to society
Mathematical
Information
Platform
5
• Creating a new paradigm in Society 5.0 (Human-centered
Society)
1.Experimental science
2. Theoretical science
3.Computational science
4.Data-intensive science(Fourth paradigm)
Aristotle’s geocentric theory (Nature philosophy)
Sir Isaac Newton (Philosophiæ Naturalis Principia
Mathematica)
Numerical calculation, Simulation
Hypothesis discovery from data
What's next?
Mathematical
Information
Platform3.Objectives and Research Area Overview (Continued)
6
4.Examples of Specific Research
1. The construction of a technology for enabling the extraction of useful information
from a limited amount of data by applying a mathematical model to the problem
we should solve (the extraction of a principle mechanism), including applications
such as the forecast and prediction detection of rare events.
2. The theory and technology for the refinement of mathematical model and the
efficient application of simulations based on the collaboration of the model- and
data driven types.
3. The creation of a new science that links mathematics, mathematical science, and
information science represented by the existing page ranks, difference privacy,
compression sensing, and RSA encryption.
4. The principles and technologies for converting natural language information and
sensing information into computable data by using the ideas in mathematics
(digitalization and symbolization of information), making data anonymous,
guaranteeing quality and trustworthiness, or sampling theory.
5. Mathematical/mathematical scientific research on security, personal information
protection, anonymity, and fairness.
Shown below are some examples of specific research.
However, the proposals we call for are not limited to the following.
Mathematical
Information
Platform
7
5.Assumed Methods to Advance Research
1. Determine which application field you aim at and choose a theme to aim at
solving social problems and/or creating new science and/or values.
2. An ideal team comprises researchers from the fields of
mathematics/mathematical science, information science, and application.
Mathematical
Information
Platform
1. Mathematicians/mathematical scientists and information scientists make a
team to conduct research, and aim at the construction of the theory and
technology that contribute to the creation of an innovative information
utilization method involving the concept of mathematics/mathematical
science.
2. Consider specific cases indicating how such technology leads to the
solution of a social problem or the creation of new values.
3. To examine social problems and the creation of new values, we think it is
efficient to have, for example, specialists from application fields such as
medical, biological, material, environmental, and energy participate as
members.
8
6.Research Projects (Year Started : 2019)Mathematical
Information
Platform
Kajiwara,
KenjiProfessor
Institute of Mathematics for Industry,
Kyushu University
ED3GE: Evolving Design and Discrete
Differential GEometry - towards
Mathematics Aided Computer Design
Kabashima,
YoshiyukiProfessor
Graduate School of Science,
The University of Tokyo
Deciphering intracellular phenomena
through information flow
Kawahara,
YoshinobuProfessor
Institute of Mathematics for Industry,
Kyushu University
Operator theoretic data analysis of
complicated dynamics and its itegerated
utilization with mathematical models
Yaguchi,
Takaharu
Associate
Professor
Graduate School of System
Informatics Department of
Computational Science,
Kobe University
Structure Preserving System Modeling
and Simulation Basis Based on
Geometric Discrete Mechanics
9
Anai, Hirokazu Senior DirectorArtificial Intelligence Laboratory,
FUJITSU LABORATORIES LTD
Iwata, Satoru Professor
Graduate School of Information Science and
Technology,
The University of Tokyo
Kokubu, Hiroshi ProfessorGraduate School of Science,
Kyoto University
Kotani, Motoko
Principal Investigator /
Professor / Executive
Vice President for
Research
Advanced Institute for Materials Research
(AIMR),
Tohoku University
Saito, Masa-Hiko DirectorCenter for Mathematical and Data Sciences,
Kobe University
Saeki, Osamu Director / ProfessorInstitute of Mathematics for Industry (IMI),
Kyushu University
Tanaka, Toshiyuki ProfessorGraduate School of Informatics,
Kyoto University
Tanabe, Takahito General ManagerOptimization Division,
NTT DATA Mathematical Systems Inc.
Hirata, Noriko ProfessorCollege of Science and Technology,
Nihon University
7.Research Area AdvisorsMathematical
Information
Platform
10
8.Cooperation with overseas and other research areas’ researchers
• In FY2020, JST has agreed with ANR to invite Japanese-French joint
research proposals, as well as regular (non-joint) CREST proposals as will
be described later.
• We host workshops with researchers from various relevant fields in and
outside the country and researchers in the fields of information and
mathematical sciences, and promote cooperation and integration between
mathematics/mathematical and information sciences.
• We aim at the multiplier effect with the research areas in the PRESTO and
ACT-X projects that have been established under the same strategy target
as this research area, and thus, co-host workshops to closely share
information among the researchers from various fields.
• Proactively promote cooperation with overseas researchers and projects
• Closely share information among the research areas under the same
strategic objective as this research area
Mathematical
Information
Platform
11
9.Considerations for application
• Determine which application field you aim at and choose a theme to aim at
solving social problems and/or creating new science and/or values.
• An ideal team comprises researchers from the fields of
mathematics/mathematical science, information science, and application.
• In your proposal, describe, as specifically as possible, what new information
utilization platform will be constructed with mathematics/mathematical
science or how mathematics/mathematical science and information science
are integrated to create innovative science and/or new values.
• Schedule(1st Term): FY2020 Existing Research Areas
- Application deadline: June 23rd, 2020 at 12:00 noon, JST
- Notification of document screening results: Late in August - Early in October
- Interview period: Middle of September - Middle of October
- Research start: December, 2020 or After
1. The upper limit of the research funds is 250 million yen (excluding
indirect expenses) for the whole research period.
2. In your proposal, describe the milestones in three years and the goal to
be accomplished in five and a half years.
Mathematical
Information
Platform
10.Japanese-French joint research proposals
• Only applications submitted to both JST and ANR will be considered for evaluation.
• Deadline for proposal submission to ANR is June 23rd, 10AM CEST.
※ Please refer to ANR’s website for the latest information.
• JST and ANR will evaluate proposals in parallel and make common funding decisions.
• Neither joint nor regular CREST proposals have an advantage during evaluation and will be
evaluated equally.
• PIs on the Japanese side may submit only one application with either a joint or regular
proposal.
• The proposal submitted to ANR cannot be changed when submitting to CREST.
• Please refer to WEB for detail information.
1. To foster Japanese-French collaboration in scientific research, JST has agreed
with French National Research Agency (Agence Nationale de la Recherche;
“ANR”) to invite Japanese-French joint research proposals, as well as regular
(non-joint) CREST proposals.
2. To apply, principal investigators (PIs) from the Japanese and French sides will
write a common joint research proposal (written in English using the Common
Application Form) and submit it to JST and ANR in parallel.
12
Mathematical
Information
Platform
13
Creating information utilization platform by integrating mathematical and information sciences, and development to society
Elucidating mathematical structures in real/virtual world objects and their utilization
Frontier of mathematics and information science *
* Additional Strategic Objective: Creating Technology for Computing Revolution for Society 5.0
Research Supervisor: Naonori Ueda (Fellow, NTT Communication Science Laboratories; Deputy Director, RIKEN Center for Advanced Intelligence Project)
Research Supervisor: Takashi Sakajo (Professor, Graduate School of Science, Kyoto University)
Research Supervisor: Kenichi Kawarabayashi (Professor/Deputy director, National Institute of Informatics)
Strategic Objective: Creating information utilization platform by integrating mathematical
and information sciences, and development to society
CREST, PRESTP, ACT-X Research Areas Collaborations
Mathematical
Information
Platform
14
JST news March 2020Mathematical
Information
Platform
Special issue: Challenges to solve social issues by integrating mathematical and information sciences
https://www.jst.go.jp/pr/jst-news/backnumber/2019/202003/index.html
AIP Network Laboratory
⚫ Japanese Government launched “Strategic Council for AI Technology” as a control tower of AI technology. The Council has come to manage
five National Research and Development Agencies that fall under the jurisdiction of the Ministry of Internal Affairs and Communications,
Ministry of Education, Culture, Sports, Science and Technology (MEXT), and Ministry of Economy, Trade and Industry.
⚫ Affiliated to the Council, MEXT launched AIP Project (Advanced Integrated Intelligence Platform Project) in FY2016. AIP Project is conducted
by AIP Center, RIKEN and AIP Network Laboratory, JST collaboratively.
Mathematical Information Platform
(Naonori Ueda)
Symbiotic Interaction
(Kenji Mase)
Artificial Intelligence
(Minoru Etoh)
Information and Future(Masataka Goto)
RIKEN Center for
Advanced Intelligence
Project (AIP)
Director: Masashi
Sugiyama
People
Results
JST AIP Network LaboratoryDirector: Katsumi Emura
*AIP(Advanced Integrated Intelligence
Platform)
Math-Structure(Takashi Sakajo)
IoT(Hideyuki Tokuda)
Humans and Interactions
(Jun Rekimoto)
Math and Info(Ken-ichi Kawarabayashi)
Human-Machine Harmonious
Collaboration (Norihiro Hagita)
Big Data Core Technology(Masaru Kitsuregawa)
Big Data Application(Yuzuru Tanaka)
Social Design(Sadao Kurohashi)
Social Information Infrastructure(Hiroto Yasuura)
Big Data Core Technology
(Masaru Kitsuregawa)
AIP Project
★
(★2020 Call for Proposal RAs)
★
★
★
Collaboration
Mathematical
Information
Platform
Appendix
Mathematical
Information
Platform
Information Science
Mathematical logic
Natural language processing
Dimensionality reduction
PageRank
Differential privacy
Software engineering
Computational science
Agent model
Network analysis
Deep learning
Security
Concealment calculationData mining
Machine learning
Graph theory
AlgorithmOptimization
Statistics
Data assimilation
Combinatorics
Topological data analysis
Compressed sensing
Geometry
Applied mathematics
Algebra
Mathematical engineering
Numerical analysis
Mathematics
Mathematical sciences
Analysis
LLL algorithm
RSA cryptosystem
Medical MaterialBiologicalSocial &
Economics
Environment
al & Energy
Solving Social Problems
Application
Fields
Mathematical Information
Platform
Science that connects
info. and math. sciences
New concept / new
theory
Based on MEXT material
Research Projects Portfolio
Kajiwara, Kenji “ED3GE: Evolving Design and Discrete Differential GEometry - towards Mathematics Aided Computer Design”Kabashima, Yoshiyuki “Deciphering intracellular phenomena through information flow”Kawahara, Yoshinobu “Operator theoretic data analysis of complicated dynamics and its itegerated utilization with mathematical models”Yaguchi, Takaharu “Structure Preserving System Modeling and Simulation Basis Based on Geometric Discrete Mechanics”
Mathematical
Information
Platform