Keywords hybrid systems, bifurcation,controlE-mail: ueta@tokushima-u.ac.jpTel. +81-88-656-7501Fax: +81-88-656-9122HP : http://risa.is.tokushima-u.ac.jp

Content: Nonlinear electric circuits including switchingdevices (e.g., multiplexer, relay, diode) are composed byboth continuous flows and discrete events. These hybridsystems often specify bifurcations of periodic attractorsand exhibit chaotic motions. We have developed anumerical analysis methodology for bifurcations in suchsystems for threshold values by computing variationsabout it. Bifurcation diagrams of periodic solutions can beobtained accurately. As an application of this technology,we demonstrate generation of periodic attractors (Fig. 1)by combining several switching manifolds, andcontrolling chaos in hybrid systems (Fig. 2) by using theperturbation of a switching threshold value. The keypoints of these trials are: (1) all required information isgiven by numerical integration of variational equations,(2) threshold perturbations are realized by less controlenergy compared with the conventional controltechnologies. We now tackle to analyze and control thehybrid system including a hysteresis characteristicswhich is also described by flows and events.

Fig. 1: Generation of attractors by threshold perturbation control

Fig. 2: Controlling chaos by threshold perturbation control

< > Content:The research focuses on a supporting framework thatcontributes motor-skill learners of repeating motions. Theconcrete target of repeating motion is “hula-hoop” at beginning.It has several characteristics such as manipulating an object,in room, simple motion.The supporting process we have in mind is shown in the left

figure. To start, the system monitors the 3D motion and createa series of motion-code. Then, it will automatically analysesthe coded motion in a wave form. There are several types ofknown wave pattern based on our initial discussions. After thisprocess, the system makes the feedback model and gives thevisible feedback image. A learner can identify which body-partand a way of movement should be improved from the currentway of movement.In fact, the supporting process described above is associated

with the assumption about the human process of motor-skilllearning. The process includes cognition, motion selection,actual motion and storing result. We have to discuss thehuman process deeply and design the framework based onthe discussion. We try to propose the designing issue and itsimplementation of a supporting system assisting the wholeprocess mentioned above.

Keywords Skill development, motor-controlE-mail: ma2 @ tokushima-u.ac.jp>Tel. +81-88-656-9804Fax: +81-88-656-9804HP : http://pub2.db.tokushima-u.ac.jp/ERD/person/73057/work-en.html

Monitoring in 3DCoding the motion

20

16

Hula-hoopRightLeft

Analysis

Feedback

Content:We can learn a CPU architecture and its behavior from

a simple CPU architecture model or a software CPUemulater. However, actual worked CPU architecture isvery complex, and have some great understanding of thewall for the learner. Therefore, we need some CPUimplementations to understand them for step-by-steplearning (Fig. 1),

In our study, we aim to develop some implementationsof CPU and associated hardware using a hardwaredescription language VHDL on a rewritable IC FPGA(Field Programmable Gate Array). The learner is able torewrite the hardware using these implementation as atemplate. We selected MIPS-I ISA(Instruction SetArchitecture) which has small number of instructions withsimple ISA. Further, we implemented the exceptionhandling and memory management unit and otherperipheral circuits on FPGAs for running a Unix-basedOS NetBSD (Fig 2).

Keywords CPU for learning, FPGA, VHDLE-mail: sano@tokushima-u.ac.jpTel. +81-88-656-7555Fax: +81-88-656-9122

Fig.1 incremental learning

Fig.2 the architecture of implemented system

Content:In this project, we have developed an IPv6 address

assignment method which is based on a popular routingprotocol OSPFv3. The developed method is free fromaddress duplication. With this method, we can reduce theamount of manual configurations of routers. Thiscontributes to remove some barriers to introduce IPv6 toa network.

Major technical fruits of this project are 1) a layout of anIPv6 host route which shows an address space anetwork can use, 2) a procedure on each router in anetwork in order to assign IPv6 addresses withoutduplication and 3) a mechanism to detect if there is aduplication of router ID.

Keywords Traffic Engineering, Configuration AutomationE-mail: ohira@tokushima-u.ac.jpTel. +81-88-656-7555Fax: +81-88-656-9122HP: http://pub2.db.tokushima-u.ac.jp/ ERD/person/305803/profile-en.html

We are researching how to improve user’s satisfaction on information technology systems using information retrieval and machine learning. Information retrieval (IR) is one of the most important research areas of industrial application. Information retrieval is an activity of obtaining information which we need from big information resources (Big Data). Machine learning (ML) is one of the hottest research areas in computer science. Machine learning is a subfield of artificial intelligence (AI) which have received a lot of attention in recent years for a technological evolution of deep learning. We are researching cutting-edge information technology systems using information retrieval and machine learning technologies. Additionally, we are aiming at generating new value on various area; information security, sports science, medical information and artificial information.

Keywords information retrieval, machine learning,natural language processingE-mail: tanioka.hiroki@tokushima-u.ac.jpTel. +81-88-656-7555Fax: +81-88-656-9122

Big Data

InformationRetrieval

encoderdecoder

Vector Space Model

Kernel Trick

Deep Learning(Autoencoder)

Information Systemqd1

d2

< > The Center for Research Administration & Collaborationwas established in April 2015 to support research andcoordinate academic-industrial collaboration, promotingthe enhancement academic research and academic-industrial collaboration fields. It comprises the IntellectualProperty and legal Division, the Innovation Division, andthe Research Administration Division. It also serves asthe general point of contact for research support andindustry, academia and government collaboration withthe university. We have achieved synergy and a nimbleorganization with the Research Strategy Office inTokushima University. As a result of that, TokushimaUniversity was ranked 5th for the institution whoseincome from patent rights increased dramatically and 11th

for the amount of income from patent rights in 1,036research institutes in Japan, according to the survey ofMEXT. Although we have limited human resource andbudget, we keep trying to make the organization functionbetter, and supporting the academic research andacademic-industrial collaboration, with our wisdom.

Keywords Industry-academia-government CollaborationE-mail:Tel. +81-88-656-7592Fax +81-88-656-7593HP

<< Center for Research Administration & Collaboration

6,670 4,636 3,518 34,457 36,791

12

6

11

14

19

02468101214161820

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

2011 2012 2013 2014 2015

Yen (thousand)Number

Yen (thousand)

Number

Number of Patent Consultation and Patent Applications (Yearly Record)

Number and Amount of Technology Transfer (Yearly Record)

42 52 64 40 51

33 49

52

45 57

75

101

116 85

108 94

83 93 88 95

0

25

50

75

100

125

150

175

2011 2012 2013 2014 2015

1 International Patent consultation

SICO, which consists of the five national universities,was launched as the government project in October 2013.We aim to utilize the intellectual properties of universitiesto stimulate research activities and the regionalcommunity by creating new business through connectinguniversities to companies that are looking for newtechnology. We have associates in each university andare taking inquiries on technology and business. Pleasefeel free to contact us.MATCI is a system intended to be used for matchinguniversity’s seeds (more than 13,500) and company’sneeds. A company can search the seeds and havecontact with a researcher through the system.(HP:https://sico-system.ccr.tokushima-u.ac.jp/search/index.html# )

Keywords Industry-academia-government CollaborationE-mail:Tel. +81-88-656-9702Fax: +81-88-656-7274HP :