Seminar Information

16th AIST Artificial Intelligence Seminar, "Monitoring the Integrity of Infrastructure Using Artificial Intelligence"

Due to the large number of people wishing to attend, we have increased the capacity of the venue.For "R&D on Learning-Based Percussive Analysis Technology," the presenter has been changed due to unavoidable circumstances.

For details, please check the program. In Japan, many bridges, tunnels, and other items of public infrastructure were constructed during an era of rapid growth; now, they are all becoming superannuated concurrently. How to maintain and manage this infrastructure has become an urgent social issue; similar issues have also arisen regarding not only large structures, but also types of infrastructure such as water pipes and power poles. Moreover, in some fields, such as power-plant and industrial infrastructure, experienced engineers are in short supply; consequently, some tasks are approaching the limits of conventional labor-intensive maintenance. Under these circumstances, there is a need to shift from after-the-fact response to pre-emptive maintenance. Accordingly, expectations are high regarding the use of artificial intelligence that can monitor large volumes of sensor data. For this seminar, we have invited Professor Takehisa Yairi (University of Tokyo), a researcher at the forefront of this field. Professor Yairi will give an overview of abnormality detection methods based on machine learning, and present examples of applications. We will also present examples of the latest research by AIST.

Title	16th AIST Artificial Intelligence Seminar, "Monitoring the Integrity of Infrastructure Using Artificial Intelligence"
Date and time	June 30, 2017 (Fri), 16:00-18:00
Reception hours	15:30 - 17:00
Venue	Meeting Room 1, 11F Annex, AIST Tokyo Waterfront, 2-4-7 Aomi, Koto-ku, Tokyo
URL	http://www.aist.go.jp/waterfront/ja/access/
Speakers	Takehisa Yairi, Associate Professor, Department of Aeronautics and Astronautics, University of Tokyo Takeshi Nagami, Senior Researcher, Computer Vision Research Group, Intelligent Systems Research Institute, AIST
Capacity	150 persons
Registration to attend	Applications will be accepted from 13:00 on June 7 (Wed). Please apply from here.
Admission	Free
Organizer	AIST Artificial Intelligence Research Center
Contact	AI Seminar Desk
Seminar details	This seminar is an activity in “Development of Core Technology for Next-Generation AI and Robots” (Next-Generation AI Field),” a project commissioned by NEDO.
Points to Note If you become unable to attend, please cancel your registration promptly, out of consideration for others who would like to attend. Please register under your real name. AIST will not use the information you send for any purpose other than seminar management. No social event is scheduled after the seminar.

Program

16:00-17:00	“Data-driven abnormality detection for large-scale artificial systems” Takehisa Yairi (Associate Professor, Department of Aeronautics and Astronautics, University of Tokyo)
	Outline Technology for monitoring the health status of large-scale artificial systems and detecting abnormalities and inefficiencies is extremely important in various sectors that support daily life, including transport, production, power, and communications. Recently, advances in measurement and communication technologies have enabled huge amounts of sensor data to be obtained from such artificial systems, and progress in machine learning research has led to the spread of various methods for modeling system behavior from such huge amounts of sensor data. Consequently, interest and expectations have grown regarding data-driven health status monitoring and abnormality detection technology. In view of this situation, the seminar will present basic approaches, typical methods, and examples regarding data-driven abnormality detection, and describe some current issues and the future outlook.
	Profile Obtained a PhD in engineering (aerospace and aeronautics) from the University of Tokyo in 1999. Following appointments as research assistant, lecturer, and associate professor at the University of Tokyo, was appointed Associate Professor, Department of Aeronautics and Astronautics, University of Tokyo in 2015. Specializes in machine learning and its application to aerospace fields. Is particularly interested in abnormality detection, dynamic system learning, and nonlinear dimensionality reduction. From 2014 to 2016, served as a director of the Japanese Society for Artificial Intelligence. Photography at seminar Photos may be shared with colleagues, etc. Seminar materials Will be posted on Web site of AIST Artificial Intelligence Research Center at a later date.
17:00-17:30	“R&D on crack detection service for road structures” Takeshi Nagami, Senior Researcher, Computer Vision Research Group, Intelligent Systems Research Institute, AIST)
	Outline Since 2014, we have been doing R&D on crack monitoring systems for road structures in NEDO’s Public Infrastructure Maintenance and Renewal Project, together with Shutoko Engineering Co., Ltd. and Tohoku University. In this work so far, we have developed high-precision crack detection technology, implemented it on a cloud platform, and constructed a system that can be used even from inspection sites. In the seminar, we will describe this initiative. For some crack detection work, we have developed and utilized learning-capable detectors using CNN. Nagami will also talk about acquisition of teaching data and implementation on cloud platforms; we would like to exchange opinions on these topics.
	Profile In 1989, graduated from Chiba University of Education, and joined the Ministry of International Trade and Industry, National Institute of Advanced Industrial Science and Technology, where he was involved in research on natural languages. In 2001, the institute became part of AIST. Is currently engaged in research on image processing and human interfaces in the Computer Vision Research Group at AIST’s Intelligent Systems Research Institute. Photography at seminar Permitted without restriction Seminar materials Will be posted on Web site of AIST Artificial Intelligence Research Center at a later date.
17:30-18:00	“R&D on learning-capable percussive analysis technology” Masaya Iwata (Chief Researcher, AI Applications Research Team, AIRC, AIST)
	Outline In the inspection of public infrastructure, visual inspection and percussive testing are widely used as primary methods. If abnormalities are detected by primary inspection, then secondary inspection is done through process observation and precision measurement. Currently, primary inspection relies on inspectors’ experience and senses. As Japan’s population ages and the number of working-age people declines, experienced inspectors have tended to decrease, and in some cases, have become in short supply. At AIST, regarding technology for SIP infrastructure maintenance, renewal, and management, we are engaged in R&D to improve the precision of current percussive testing, and quantify the degree of abnormalities using machine learning, without relying on inspectors’ senses, to prevent inconsistencies and errors. In the seminar, we will present the latest results, including actual examples.
	Profile 1993: Graduated from University of Osaka, majoring in applied physics. 1993-1994: Visiting Researcher at ONERA Toulouse, France 1994: Joined Electronics Research Institute (OK??) (currently AIST). Is currently Chief Researcher, AI Applications Research Team at AIRC. Photography at seminar Because the presentation includes unpublished results, please refrain from photography during the seminar. Seminar materials Not scheduled for public release