## 20Nx Ȋw mwʘ_$T ## 0761029 ec@Dm u{bg̐lɑ΂񎦂̌v @{_ł́A{bg̐lɑ΂񎦂̗lXȎ@ɂďqׂB @ʂɃ{bg͐l̑ɍƂsƂ҂B̂߂ɂ́A{bg͐lƂ̑ݍpɂčƂKvBK؂ȋs邽߂ɂ́AƂɊւ񂪐lɑ΂ēK؂ɒ񎦂KvB lƑݍp邱ƂōƂs{bgƂāAu샍{bgA^{bgA퐶x{bǧsĂĂB̃{bgɋ߂郆[UC^tF[XɂāA{bglւ̏񎦂ɂĕKvȋ@\ĂB @u샍{bg̐lɑ΂񎦎@ƂāAumIj^OvƂlĂB́AuƃVXeɂAƊĎ̂߂̐VȒmI@\̈ƍl̂łB{_ŒĂ邻̋@\Ƃ́Auƃ{bgVXepƎsɂāA҂̃{bgɑ΂Ǝw߃R}hɑ΂āA{bgΉƎssȂƂƂɁAX̍ƂɓKĎ̑҂ւ̒񎦂̒mIȐsȂƂ̂łB @^{bg̐lɑ΂񎦎@ƂāAuvWFNV@\vĂBvWFNV@\͏e[uʁAǂȂǂ̃{bgɂƂs̂̂ɁArfIvWFN^[Ń{bglւ̃bZ[WƂȂf𓊉e邱Ƃɂč\Be摜ɂāA{bgƓɂčƂčsĺAƂɊւ{bgƋL邱ƂłBefƂāÃ{bg̃V~[V摜pB @퐶x{bg̐lɑ΂񎦎@ƂāA{bg̎s\ȍƃf̃Xg̒񎦂ƐlɂIƂ@ĂBƃf̑ǏA{bǵA̎sɕKvȏlɖ₢킹B̎ɑ΂铚́ARs[^ʏŎƊ̊Ď摜Ń{bg̗vɏ]ă}EXŐ̓_NbNƂȒPȂ̂ɏWłBĎ摜̓XeIJŎBeĂAʂ̋摜vĂB{bǵA摜ŃNbNꂽ_̋f[^𗘗pāAƂ𐋍s邱ƂłB @ŌɈȏ̎@𓝍A퐶ɂĐlƋA郍{bg̏Ƃ̊J̓W]ɂďqׂB ## 0561018  Studies on minimum throughput assurance service in provider provisioned virtual private networks (voC_Ǘ^VPNɂŒшۏ؃T[rXɊւ錤) @C^[lbg͊ɎЉIȏՂƂĂ̒nʂmĂAN𑝂Ƃɗp҂ƃgtBbNʂ͑ĂBÁAAdvȋƖȂǂłC^[lbg̗֐͊pA荂M߂BC^[lbg̓xXgGtH[g^̃T[rXƒmĂAp҂̑ш͕ۏ؂ꂸts̑ш̊mۂdvƂȂB]Aш̊mۂKvƂȂf[^unőMꍇɂ͐ppĂAp̐ݒup͍AC^[lbgŉzIɐpԂ\zVPNiVirtual Private NetworkjڂĂBߔNł́AvoC_̏LԂpA҂ւ̊шۏ؂voC_Ǘ^VPNڂĂB @{_͉p4͂琬B1͂ł́A܂C^[lbgł̑шmۂɊւۑBvoC_Ǘ^VPN̗֐ɂďqׁAŒшۏ؃T[rX̏dvƉۑɂďqׂBŌɁAŒшۏ؃T[rX̎ɕKvƂȂш抄ƃvrWjOASYɂďqׂB @2͂ɂăz[Xш抄ĂBvoC_͉҂Ƃ̌_Œ߂ꂽvш𖞂ш񋟂Ȃ΂ȂȂÄŗLȎłшǂpȂ΂ȂȂBĕł͌qtB[hobN^ш抄VPNz[Xf𓝍邱ƂŁAv𖞂шǂpBO҂̓lbg[N̏ʒmpPbgɂĔcA{glbNN̗p\шɉđш蓖ĂB҂VPNŏ]pĂΈ̋_Ԃ̃RlNVW񂷂邱ƂŌǂш𗘗pBĕ̕]ł́AV~[VɂʓIȌʂɁA̗LB @3͂ł͔v@pvrWjOASYĂB2͂ŏqׂtB[hobN^ш抄ł͏d݂ƂlpāA̒l̔ɊÂđш̊sBĕł́AVPN҂̗vшƃvoC_ۗL郊Nш񂩂Avш𖞂ш旘pő剻d݂肷B @4͂ɂĘ_̂܂Ƃ߂sBȏAvoC_Ǘ^VPNɂčŒшۏ؃T[rX邽߂̑ш抄ƃvrWjOASYɊւāA̗L𖾂炩ɂB̒mAVPNшۏ؂߂镪ɂāAŒшۏ؃T[rX̎\ƂȂB ## 0161202 @ uσ}pXLϒPSK-VPɊւ錤v @fW^ړ̒ʐMɂẮA}pXg݂Ɋ邱ƂŒʐMi򉻂BړƋɁAepX̊ŒUƈʑ̕ω𔺂tF[WO𐶂邪ADopplerL肪xɔ䂵ĖłȂ炢̍tF[WOɂȂƁAʑUŌ̕Ȃ蒘ʐMi򉻂BAŏV{ZȂAΓIɃ}pXg̒xL肪łȂI𐫃tF[WOƂȂƁAMш̎g̍rɂgcɂV{Ԋ}ɑ債ȀꍇʐMi򉻂B @LxLւ̑΍ƂāA܂XyNggUnZpAčXȂ鍂ցA𑽐̃TuLAɐU蕪OFDMoꂵBOFDḾAV{xIɗƂƂŁAxL̉e{IɒጸłB́AGHzȏ̃}CNgт⏀~E~gтpA܂AS^]x̍ړԗւ̐vmȒʐMɌĈiƒ藦vADopplerLł̗򉻂łȂꍇ\zBOFDḾAV{xTuLAԊŕsɂȂ₷AxLDopplerLւ̑ϐ̓g[hItƂȂBāAHւ̐KKvƂApCbg\ɂẮAHϓւ̒Ǐ]sǂ򉻂𐶂B @{ł́AϓHւ̐KsvōtF[WOɖ{IɋAVȑσ}pXϒƂāAPSK{Ɉʑ璷̂PSK-VPiPSK with Varied PhasejƑ̂ĂB{́At璷ɂāAH̕ϓɑ΂ĈقȂω镡ނ̗LȌgo͂AgtB^ō邱ƂŁA}pXŃ_Co[VʂɂϋɓIȉP̂łB̍ڂƏ璷̓őɂFXȃoG[V݂ĂĂ邪ATāAxLɑ΂AP݂rIBσ}pXϒQɂƂāAexԍ傫邱Ƃ́A璷ɂ鏊vш敝̍L}邱ƂƕœKp͈͂LŏdvȉۑƂȂB @{ł́A܂AĂPSK-VPɊւA̓P@\𖾂炩ɂA璷ʑgƌ藦̊֌W𖾂čœKȈʑg߂BāAœKp^肵X̊{𖾂炩ɂAexԍƏvш敝̓_ŁAeQ̒ŗDʂł邱ƂmFBH͂ƕāAłPSK-VP@]ѓʂČ؂Bɂ́A]ʂĕ߁AɌpWJуVXeB̓Iɂ́AUAeiɂ镡ǓMƑgݍ킹Aړԗւ̘HԊԒʐMɂāAsRZq킹ĈӐ}IȖGAsVXelĂB̃VXevZ@V~[Vю@]ɂĖ炩ɂAɁAtB[hɂČ؊mFAPSK-VPpYVXe̓EDʐ𖾂炩ɂB @{ɂAtF[WOɋAeQ̒ł͋exԍƏvш敝̓_ŗDʂȁAVȑσ}pXϒPSK-VP̕ZpmBɁAY̓čړԗւ̐vmȒʐMłAӐ}IȖGAsVXeыZpmApWJւ̉\B ## 0561020 R@i uOn the Minimum Weight of Simple Full-length Array LDPC codesv @͎ĜʐMHĎMꂽMV{Ɍ肪Ă邩ۂA肪oΌ߂ꂽ葱ɂđM𐄒肷ZpłBMɌ肪Ă邩ۂo邽߂ɂ́AʐMH̎GɂđM̂̃rbg]Ă̑MƓɂȂĂ܂ȂƂKvłBقȂ̑Mōł߂̂̋ŏƂBŕԂ̋Ƃ͑قȂrbg̐B̍ŏ́A̐̕$Ŕɏdvȃp[^łB
@ᖧxpeB(LDPC)(Low Density Parity Check code)͑aȌsŒꂽŁA1962NGallagerɂĒĂꂽBᖧxpeBsƂ́AŝقƂǂ̐0łA1̐ɑaȍsłBLDPC ́AShannonEɔ鍂\͂ƂĒڂĂAfBW^qKiDVB-S2ւ̗̍pIEEE802.16eAIEEE802.3(10G BASE-T)ȂǂŎpĂB ALDPC̐\͊SɒʓIȕ]ȂĂ킯ł͂ȂAႦ΍ŏ̕]͒萫IȐ\]ivI@ɂ鐫\]j@p邱ƂقƂǂłB́ALDPC̃peBs񂪏]AvZ@pă_ ɐ邱ƂƂɋNB AߔN[LDPĈ悤ɑ㐔I̍\@pLDPC̍\@ĂȂ㐔I\𗘗p̐\]@ĂȂǁAʓIȐ\]ssĂB Simple full-length A[^ LDPC (SFA-LDPC)́A㐔Iɍ\ꂽ LDPC̕NXłAPŐwI\B{_ł́ASFA-LDPC̍ŏd݂ɂċc_B SFA-LDPC̍ŏd݂, Yang, Mittelholzerɂ㐔IȃAv[猤ȂĂA^̍ŏd݂́Aȃp[^ŋK肳ꂽ̃NXɂĒmĂ݂̂łBp[^ 傫ȕɂẮA̍ŏd݂̉E, Ec_Ă邪A㉺E̊Ԃɂ͑傫Ȋu肪Amȍŏd݂̉𖾂ɂ͂ĂȂB ɑ㐔IȃAv[ɗ̂ł́AK肷p[^傫Ȃɏ]AeׂɌׂ̃p^[Iɑ邽߁Aŏd݂肷邱Ƃ͍łƗ\B ŁA{eł͂܂SFA-LDPC̏ElZ@ĂB镄̒ɓ̏d݂ꂪ݂邩ۂɂ́A̓̏d݂xNgS邱Ƃɂ\ł邪Â悤ȃAv[ׂ͌xNg̐cɂȂ邽ߌIł͂ȂB{ẽAv[́A㐔IȃAv[ɌvZ@pAL̑㐔IȌ̎葱̂łB̕ɂAŏd݂mł SFA-LDPCɂāA̐mȒl𖾂炩ɂ邱ƂłB
@{ě㔼ł͌s̗d݂4̂ƂA5̂ƂSFA-LDPC̍ŏd݂̏E̕]̑㐔Iؖ^BǑʂڍׂɕ͂邱ƂɂASFA-LDPCK肷2̃p[^̂̂ЂƂ(̗d݂𐧌䂷p[^)Œ肵ꍇAЕ̃p[^ςĂ̂̕ȂɕK܂܂镄̃p^[łB̃p^[㐔Iɏؖ邱ƂɂAŏd݂̏̕]B̍ŏd݂̏̕]@pAs̗d݂4̂ƂA5̂ƂSFA-LDPC̍ŏd݂Aꂼ10, 12ł邱ƂB݂͌܂łɒmĂŏd݂̏̕]啝ɉP]łB

## 0561211 Fawnizu Azmadi Hussin uStudies on Core-Based Testing of System-on-Chips Using Functional Bus and Network-on-Chip Interconnectsv

@The tests of a complex system such as a microprocessor-based system-on-chip (SoC) or a network-on-chip (NoC) are difficult and expensive. In this thesis, we propose three core-based test methods that reuse the existing functional interconnects\a flat bus, hierarchical buses of multiprocessor SoCs (MPSoC), and a NoC\in order to avoid the silicon area cost of a dedicated test access mechanism (TAM). However, the use of functional interconnects as functional TAMs introduces several new problems.
@During tests the interconnects\including the bus arbitrator, the bus bridges, and the NoC routers\operate in the functional mode to transport the test stimuli and responses, while the core under tests (CUT) operate in the test mode. Second, the test data is transported to the CUT through the functional bus, and not directly to the test port. herefore, special core test wrappers that can provide the necessary control signals required by the different functional interconnect are proposed. We developed two types of wrappers, one buffer-based wrapper for the bus-based systems and another pair of complementary wrappers for the NoC-based systems.
Using the core test wrappers, we propose test scheduling schemes for the three functionally different types of interconnects. The test scheduling scheme for a flat bus is developed based on an efficient packet scheduling scheme that minimizes both the buffer sizes and the test time under a power constraint. The scheduling scheme is then extended to take advantage of the hierarchical bus architecture of the MPSoC systems. The third test scheduling scheme based on the bandwidth sharing is developed specifically for the NoC-based systems. The test scheduling is performed under the objective of co-optimizing the wrapper area cost and the resulting test application time using the two complementary NoC wrappers.
@For each of the proposed methodology for the three types of SoC architecture, we conducted a thorough experimental evaluation in order to verify its effectiveness compared to other methods.

@

## 0261005 }y@ uq[}mCh{bĝ߂̃}[_C^NVɊւ錤 Multi-modal Interaction System for Humanoid Robotsv

@ߔNAł̎gpѐlԂƂ̋ړIƂq[}mCh{bǧɍsĂAlƂ̋ƂAl̐Ԃɂړ\ɂȂB{bg̋@\EOςlԂɋ߂̂ɕω邱ƂŁAlƂ̃R~jP[VԂϗe邪ARȃR~jP[VƌĂׂ͖̂ĂȂB {̖ړÍAlƃq[}mCh{bgƂ̃C^NV~ɍsƂłVXe\zE]鎖łBꂽVXéAvɂACR^NgƁAbAFGWɂ鉹FƂ}[_ȃC^NV@\ƂB
@3 ނ̃C^NV{bgVXe̊JE^pʂāAACR^Ng͂߂Ƃ񌾌Ɖlƃq[}mCh{bgƂ̃R~jP[Vɗp鎖̗LmFB

@

## 0561207 ޖؖ@G uԉ@pIȉR[pX\zt[[Nv

@B}Rtfɂ鉹fx[XƂAFAvP[V̊JɂāA^XNˑ̖ɂA\ȉf쐬邽߂ɂ́AړI^XNɓf[^̎WK{ƂĂBȂȂAf[^̎W͖cȃRXgƎԂKvƂAFAvP[V̊JRXgŜĂB
@{_́Aȏ̉ۑ𓥂܂AԂ̉@AR[pX̖ړI^XNɑ΂ėp̔@A^b҂̗\Iɂ鉹R[pX\z̒RXgɊւ@ĂB܂AԂ̉\Ƃ邱ƂŁA^XNԁC^XN̉Iϓ𒼊Iɔcł邱ƂmFA]̉@ƔrɂĂA̗DʐBėp̔ɂẮA]́AR[pXQAIɓ̋߂f[^I邱Ƃ͏oĂAI쐬ꂽf̐\ۏ؂邱ƂłBĂėp@p邱ƂŁAړI^XNƊ^XNƂ̉ԏ̕z̏dȂArIxɁAėpcł邱ƂmFBɁAWΏۂ̌b҂̏ʉf[^A@pĉF\Ɋ^b҂\IAIꂽb҂̉f[^W@ĂBł́A]ׂ̖ɘb҂I菇ƔrA啝ȃRXg팸\ƂȂ邱Ƃ؂B

@

## 0761003 s@ Noise Reduction Front-End for Robust Speech Recognition using Multi-Channel Signals and Harmonic Structures@ i}lMƒg\𗘗poXgF̂߂̎GtgGhj

@ʂɉF͔wiG̉e󂯂₷BŁA퐶ɂČoǴA푽lłBႦ΁Aԓł́AsG󒲑̑ɁAȂ㕔Ȃ̓҂̔AWIȂǃI[fBI@킩̍ĐACp[쉹AԒʉ߉Ȃǂ̔G݂B܂AقڒłĂAEJsȂǂ̑sG̉ł́AɒႢSNƂȂAF̐x͑傫򉻂B
@{_ł́A瑽lȎGɑΏ邽߂ɁAR̎@āE؂B1ڂ̎@́Avt@CtBbeBO(PF)ƖtV}CNtHAC̋ZpłB鉹̊pxʃp[ziϑvt@CjɒڂAm̃ev[gvt@Cɐ邱ƂɂAړI̐M𒊏oB̋ZṕAʒuƂėp邱ƂłBQڂ̎@́ASSEC (Simultaneous adaptation of spectral Subtraction and Echo Cancellation)ƖtVGR[LZ̋ZpłB́AsGłƂ̂ƂɁAXyNgTugNṼGR[LZ̓KƒG̐Ƃ𓯎ɍsBɂAsɃI[fBIĐꑱĂƂ󋵂łsGƃGR[iĐj̗𐄒肵邱ƂłBRڂ̎@́ALPE (Local Peak Enhancement)Ɩt@łB́AԂ̍EJsȂǂ̍LшɍLGɖꂩAg\𗘗pċB]Zpł́Amȃsbg̐ƗLE肪KvŁAł͂΂΂̐xƂȂAĎ@ł́Aϑp[XyNĝ̂AڃtB^݌v邱ƂɂAւ̈ˑrB

@

## 0561205 ؈@S Domain Adaptation of Statistical Word Segmentation System (vIPꕪ̕K@)

@ĂȂ{⒆ł͕̒Pւ͎̕łȂB̌̒PꕪɂĂ͓vIȎ@KpA̗LĂBAۂ̉pł͒PꕪwKf[^ƈقȂ镪ւ̓KpɌb╶̂̈Ⴂɂ鐫\̒ቺɉۑƂȂĂBŊwʘ_ł͒Pꕪ̕KɗLȈȉ2̎@ĂB
@1_ڂ́Äꕔɂ̂ݒPꋫEt^iIAme[Vj@łB̏dvƎv镔݂̂ɏWł邱ƂɂAV̊wKf[^̍쐬IɂȂB{ł͕ɕIɃAme[Vt^ꂽf[^pďtm(CRF)wK@ĂBCRF͒PꕪɓKvfł邱ƂmĂ邪ACRF ̊wK@ł͕ŜAme[Vꂽf[^z肵ĂBŁAӖޓxړI֐ɂ邱ƂŕIAme[VpCRFwK@ĂB
@2_ڂ́AK敪ł̐\ő剻悤ȏdvxd݂Kf[^ɕt^@łB{ł͊wKf[^ƃeXgf[^̖̓͂xTv璼ڐ肷@ĂBĎ@̌vZʂ̓eXgTvƂ͂قړƗȂ߁AĎ@̓Ame[VĂȂKf[^ʂɓ\łPꕪɓKĂB
@{ł́AL̒Ď@ɂ蓝vIȒPꕪ̓Kł̐\؂vZ@̌ʂB

## 0661201 rJ@F Video Mosaicing Based on Structure from Motion for Geometric Distortion-Free Document Digitization (􉽘ĉ݂ȂfW^CWOړIƂ摜̎OɂrfIUCLO)

@gѓdbȂǂ̃Jtgы@pĕfW^CWOsȂ@ƂāArfIUCLOɊւ錤ɍsȂĂBrfIUCLOł́AʑŜ𓮉fƂĊgBeAet[摜̈ʒu킹sȂƂŁAUCN摜ƌĂ΂ꖇ̍𑜓x摜𐶐BȂAʂɁAUCN摜ɂ͎ɏqׂQނ̊􉽘c݂Ƃ肪BPڂ́AJʂƐ΂ĂȂꍇɔˉec݂łBQڂ́AJ̖{̂悤ɎʂpȂĂꍇɔȖʘc݂łB
@{ł́Å􉽘ĉ݂ȂUCN摜𐶐@ƂāAQ̃rfIUCLO@ĂB 1ڂ́AΏۂ𕽖ʂɌ肵A摜̎OɂĐ肳ꂽet[̃JOp[^pāAzIɐΉAˉeĉ݂Ȃ𑜃UCN摜𐶐@łBQڂ́AJOp[^ƕĎʂ̎O𐄒肷悤ɏL@g邱ƂŁApȎʂzIɕʂɓWJAȖʘĉ݂ȂUCN摜𐶐@łB
@Ď@ɊÂVXeJAʁAȖʂ̂Qނ̎ʂɑ΂ĎsȂBUCN摜̘c݂ʓIɕ͂ʁAʁAȖʂ̗ɂāA􉽘c݂Ă邱ƂmFB

## 0461039 @~R uThe ability of quantum information processing under the resource-restricted circumstancesv

@ @This dissertation provides the studies on quantum information processing, especially under the circumstances that the computational resources are restricted. Quantum computing is a new computational paradigm based on the quantum mechanics. It has excellent potential abilities of information processing compared to traditional computing called classical computing. However, ideal quantum computers would not be implemented under the current technology and the various computational restrictions are considered to be imposed on the actual quantum computers. Thus, it is quite important to clarify the ability of quantum computing under such restricted circumstances. The main results of this dissertation are as follows.
@First, the recognition ability of the quantum computational model with the memory restricted to a stack, quantum pushdown automata, is compared with that of the deterministic pushdown automata in a deterministic scene. In the computational model theory, the relationship between the recognition abilities of the quantum and classical automata is still an open problem and some negative results which show that the ability of the quantum computational model is weaker than tha classical counterpart are proposed. Thus, it is not obvious that the recognition ability of quantum automata is superior. The dissertation shows that quantum pushdown automata can solve a certain problem deterministically which cannot be solved by deterministic pushdown automata. The modified generalized Ogdenfs lemma is utilized to show that deterministic automata cannot solve the problem. This implies that quantum pushdown automata can be more powerful than classical counterparts.
@Second, a new quantum secure direct communication protocol is proposed. Most of the current quantum secure direct communication schemes use the brilliant resource unique to the quantum information processing, quantum entanglement, which requires the extremely delicate handling. In contrast, the proposed protocol employs no entanglement resource at all. Thus it can be said that the feasibility of implementation of this protocol is higher than the other proposals under the current technology. The proposed protocol can send quantum information as well as classical information. Thus, in order to discuss the security of the proposed protocol, a new criterion is needed which can measure the amount of quantum information. This dissertation introduces a new criterion that is based on fidelity of quantum states, and it is shown that the proposed protocol satisfies it against the man-in-the-middle attack.

## 0561202 @ uSOSœKpނԌn̈͂Ƃ̉pɊւ錤v

@ʐMɂf[^]MAȊwɂ锽ȂǁAX̐g̉ɂ͌ɑ΂錋ʂxČ悤Ȍ(ނ) ݂BނԂ̓tB[hobNns艻VXe̐\򉻂邽߁A ̈萫͂␧n݌vɊւ錤ɍsĂB
@ߔNł́ARs[^̍\_@ɑ\鐔lvZ@̓oɂAlvɑ΂鍂@𗘗p萫͂␧n݌v@ڂ𗁂тĂBȂA̔nƂĕ\ނԌnɑ΂āAL̏nƐea̍ss(LMI) pĂ邽߁Aʂɂ̎@͏\AKv\Ƃ̊u肪傫ێIȉƂė^B
@Ŗ{_ł́AނԂ܂ތnɑ΂āAێ琫̒Ⴂ͎@^B܂A𐧌n̕]@ƂĉpB
@܂AނԂ̒ԓIɕωȂꍇlAcomplete-type Lyapunov-Krasovskii Ċ֐ƓaiSum Of Squares) ̂߂̐l̓c[SOSTOOLS p萫͂ɂčl@Bcomplete-type Lyapunov-Krasovskii Ċ֐̑ݐ́AނԌn̈萫ƕKv\̊֌Wɂ邱ƂmĂ邪ALMI ł͕ێ琫Ȃ邽߁A萫ۂނԂ͈̔͂ςĂ܂Bł͓aƂď̗LAlvZɂmFAێ琫̒ጸƌvZԂ̃g[hItɊւČB
@ɁAԈˑނԌnƌĂ΂nɑ΂ߎf\Ƃ͎̉@ĂBނԂ̒ԂɈˑĕωn͏]AނԂ̕ϓsmƂđAɑ΂ăoXg䗝_KpƂAv[̂ĂAmIȋnsmnƂăfƌς肪ێIƂȂB ̂ߒĖ@ł́A UԔԕƂďԈˑނԌnߎÄ萫͂ASOSTOOLS pʃAvmt֐̒TɋABɂsmnƂẴfOɔׁA 萳mȓ̋Lq͂\ɂȂAێ琫ጸłB
@ŌɁASOS 萫͂pāAnɑ΂]@ĂBł̓rWAEtB[hobNnނԌn̈ƂĎグAۂɎړIƂĐ݌vꂽn̂ނԂɑ΂]TςB

## 0361027 @P@ u{MՂɌl̒oƃICMҏƍɊւ錤 Studies on Extraction of Individuality Appeared in Japanese Handwriting and Online Writer Verificationv

@ICMҏƍ͕̌MՂɌ肵̂قƂǂŁAʓIȕΏۂƂĂȂAMՂ̂ǂ̗lȓɌl̂Ƃ\ȂĂȂAIdvꓮIƐÓI𑍍IɗpĐ^U𔻒肷Ƃϓ_ɖRƂ肪BŁA{ł́AʓIȊΏۂƂĂǂ̗lȓlɌl邩܂炩ɂǍʂɊÂICMҏƍ@Ăɂ肻̗L]邱ƂړIƂB
@܂Aɋ߂Xg[N\J^JiΏۂɂė\sBMՂɌl̂͌lKĂ鏑ʋZ\ɍ邽߂ƍlAʋZ\ɊÂlAlɌl邱Ƃ𖾂炩ɂBɁAMҏƍѕMҎʎsAolpĕMҎʁEƍ\ł邱ƂB
@ɁAɂČBʋZ\ɊÂl͕MՊӒ̌ڂƗގĂ邽߁AlMՊӒ̒mɊÂĐђǉBol𕪐͂邱ƂŌl₷l𖾂炩ɂAl̕MҎʗ͂̕]sMҏƍ̉\B
@ɁAp҂VXe̒񎦂pX[h͂邱ƂŌlƍsVXez肵Â߂̕Mҏƍ@ĂBl̎ނƂɖ{lԂƑlԂ̋z߁AɊÂēl̑IsBɁA^MՂ̊el͎QƕMՂ̕ϒlɋ߂̂Â͏ȂƂlɊÂAel̎QƕMՂ̕ϒl̕΍̓xzɂʊĂA̗LMҏƍsƂɂ薾炩ɂB
@ŌɁAĂMҏƍ@AUMՂ̔rɑ΂ėLł邩ÅpĂLł邩ǂB\ɌPꂽUMՂWĎsĎ@UMՂɑ΂ĂLł邱ƁAɁÅɑ΂ĂLł邱ƂB

## 0461019 ˓c@S Quantitative Studies on the Systems of Rho-Family Small GTPases and Regulation of Cellular Morphodynamics (Rhot@~[ᕪqGTP^pNƍזEԐɊւʓIȌ)

@IłȂP퐫ۂ@\́A̖{IȐ̈łB{ł͍זEԂ̕ωƁA𐧌䂷Rhot@~[ᕪqʂf^pNVOiɒڂAɂזEǑIȕωɑ΂āAVXe_ɑΉP퐫ۂזE@\ɂċc_BɁA̗lȓIȐۂ𒲂ׂ邽߂ɕKvȕ@_ACuZC[WOf[^̐Vʉ@ĂB܂A̎@ۂ̃f[^ɓKp邱Ƃł̎pƗLmFBŋc_@́ACuZC[WOZp̔WƁAvZ@\̌ɂ菉߂Ďꂽ̂łAȊw̕@_ƂĐVw̂ƌBʂȊϓ_猾΁AʓIȉ͂萫IȒm𓾂邱ƂړIƂ@ɂĂ̋c_łB
@\ł͂܂A̐_oזEɑ\悤ȁAזEԂ̐䂪A̂ɂƂĂǂ̗lɏdvł邩TςA܂ł̌Ŗ炩ɂȂƂƁA܂ł̌@Œׂ邱ƂIȐۂɂďqׂBɁA̗lȓIȌۂ߂̕@_ƁA֘A錤ɂĐGBɁAזEԂ̐ɐ[ւĂ邱ƂmĂRhot@~[ᕪqʂf^pNVOiƁA̓IȐ𒲂ׂ邽߂̐f\zAvZ@V~[VɂďqׂB܂AV~[Vɂ蓾ꂽʂAG^pNԂ̑ݍpɂĎזEԂ̕ωƂ̋@\IӋɂċc_BāAvZ@pCuZC[WOf[^̒ʉ͂ɂċc_Bł͌ԕω̎n͂ۂɒʂTςA邽߂ɊJ@łEdge Evolution Tracking (EET)ƁAԕὡ|{זEɂARhot@~[ᕪqG^pN̎ԊzL^uGlM[ړ(FRET)摜ɓKpʂB܂Aꂽʃf[^瓱ꂽ萫IȐƁAɍlזE@\ɂẲc_BɁAɍ\zfԓIɊgAœꂽʂƐf̌ʂ𓝈ꂵĉ߂邱ƂōזEԐ̑S̑B܂AœꂽʂɁA^pNVOȋݍpƍזEԂ̕ωԂ̕ωۂA̑傫ȕωɑΉAŜƂčP퐫ۂ@\ɂċc_B
@ŌɁA{œꂽʂ𑍊AזEԕω̐Ƃɂ߂ēIȌۂɂĂ̍l@qׂƂƂɁA̗lȐۂ𒲂ׂ邽߂̕@ƂėL͂łƍl鐔I@ƒʓIȎ@̍̓WJɂďqׂB

## 0661029 腁@z Characterizing, Deriving and Validating Safety Properties of Integrated Services in Home Network System (z[lbg[NVXeɂƓdAgT[rẌS̒AoA؂Ɋւ錤)

@The home network system (HNS, for short) is a system consisting of multiple networked household appliances and sensors. The great advantage of the HNS lies in the flexible integration of different home appliances through the network. The integration achieves value-added services. In developing and providing the HNS integrated services, the service provider must guarantee that the service is safe for inhabitants, house properties and their surrounding environment. Assuring safety is a crucial issue to guarantee a high quality of life in smart home.
@In this research, we proposed a total framework for characterizing, deriving and validating the safety properties within the integrated service of emerging home network system, consisting of the following three contributions.
@The first contribution is to propose a way to formalize the safety of the HNS integrated services, considering the nature of the HNS and integrated services. The safety was defined as (1) local safety which is the safety to be ensured by the safety instructions of individual appliances, (2) global safety which is specified over multiple appliances as required properties of an integrated service, and (3) environment safety which is prescribed as residential rules in the home and surrounding environment. We also formulated the safety validation problem based on the safety classification.
@The second contribution is to propose a requirement-engineering approach for deriving the high quality verifiable safety properties systematically. Specifically, we proposed a new hazard analysis model, called HNS Hazard Analysis Model (HNS-HAM, for short), which consists four levels. By constructing HNS-HAM and investigating potential hazards within the given HNS model, the safety properties and their responsible operations can be derived. Moreover, to enhance the reusability of the HNS-HAM, we also have proposed the notion of the hazard template which can be reused for various kinds of the HNS objects for the common hazard context.
@The third contribution is to propose a method for validating the safety properties by using the technique of Design by Contract (DbC, for short). The key idea is to cope with the safety validation problem as a set of DbC contracts between calling and callee objects. The contracts can be verified during the runtime of the program under testing. During the execution, if a contract is broken, an exception is thrown. Thus, the safety validation problem can be reduced to the testing of the HNS implementations.
@We believe that the proposed total framework can help the HNS developers significantly in designing and implementing safe HNS solutions.

## 0661006 c@׍G M_D摤hCoʉߍsf̍\zƏo̔V~[V

@ߔNAʎ̌ьʎ̎SҐƂɌXɂ邪AˑRƂčxɂAʎ̂ɑ΂΍􂪋߂ĂBʎ̂̒łAo̂́AǓˎ̂Ɏ2ԖڂɑĂAŜȂǂ̏dȎɌqP[XALȑ΍􂪋߂ĂB܂,o̖̂80߂閳M_ł̎̌̑̓q[}G[łƌĂB{ł́AM_ł̏o̂ɏœ_𓖂ĂāAq[}G[lԂ̋@\̗򉻂̔ɋyڂeʓIɕ]Rs[^V~[Vɂčl@B
@ɁAߋ̏o̂͂̕猩ꂽT^IȐ̃q[}G[hCofɑgݍ񂾁B̃fpĂ낢ȏŌ_ʉ߃V~[VJԂƂɂāAeq[}G[o̂уqnbgɎ銄𒲂ׂ邱ƂłB̌,_ꎞ~WƂq[}G[͑̃q[}G[ɔׂďo̂уqnbgɎ銄ƂmFłBŌ,\zʃV~[VVXep,o̑΍ƂāAꎞ~OŃhCoɌxoA[VXěʂ]BɃhCo̐SgԂቺăA[VXeւ̔xꂪ邱ƂlƂAꎞ~Œ~邱ƂCÂA[Kvł邱ƂmFłB
@{ŒĂʃV~[VVXegƂɂāAhCo̔Fm񏈗vZXŋN邢낢ȃG[@\򉻂̂qnbg̔ɋyڂe]łƂƂɁA낢Ȏ̑΍􂪃hCo̍sǂ̂悤ɕςÂqnbg̔h~ɂǂ̒x̂]邱ƂłB

## 0561209 qc@Fk lbg[N^EFAugɂ钍ߏ̋Lƒ񎦂Ɋւ錤

@{_́Albg[Nɂ̊lyђ񋟂\ȃEFAugɂ钍ߏ̋Lƒ񎦂Ɋւ錤łB
@EFAuǵAĎgpEFAuRs[^pČɉzd􂵂Ē񎦂ZpłAɂ茻ɑ݂IuWFNgɊւ钍ߏ𒼊Iɒ񎦂邱Ƃ\łBʓIȃEFAug̗pz肵ꍇA߂̓eΏە̂̈ʒu͎Ԍo߂ƂƂɕωꍇl邽߁Aߏ̍XV͏dvȉۑłBɁAIȏ񎦂sɂ́A߂̒񎦕@̍Hvir[}lWgjKvłB{ł́ARɈړ\ȃEFAuRs[^̃[Ulbg[N𗘗p\ł邱ƂOƂAlbg[NLꂽߏ̊lɂŐV̒ߏ̒񎦂̎Ayуr[}lWgɂ钍߂̗x̌ړIƂB
@]̃EFAugVXéAߏ[UEFAuRs[^ɂ炩ߕێ̂قƂǂłAߏ̒ǉEXVsƂłƂ肪Bɑ΂Aߏ̒ǉEXVIɍslbg[NL^̒ߏf[^x[X\zBɂAߏ̒񋟎҂́CT[õf[^x[XXV邱ƂŁAɍŐV̒ߏ[Uɒ񋟂ł邱ƁAEFAuRs[^̃[ÚAlbg[NčŐV̒ߏlA{ł邱ƂB܂AԂňړ镨̂ւ̒ߕtANCAgT[oʐMP2PʐM𕹗pnCubhP2PɂB
@ɁA߂̗x̌̂߂ɁAEFAugVXẽ[U̎Eɑ݂钍ߑΏە̂̑ݗ̈𐄒肵Ax̒ቺɌqvfyieB֐邱ƂŁAyieB̍ŏɂr[}lWg@ĂBĎ@ɊÂVXe쐬AEFAuRs[^̃[U𒍎ߑΏۂƂɂAr[}lWg𗘗pߕt̗LmFB

## 0661018 KI sKxllbg[Nn̐݌vɊւ錤

@ʏNCSɂẮAx͖m̎ϗvfƂăf邱ƂBApPbg̑MɃ^CX^vtAMɒxԂ߂@mĂB̌v@ɂAΏۂƂNCŚAxԂXCbOMƂȂnƂĒ莮ł邽߁Aؑւ̐݌v@Kp邱ƂłB
@ŁA{_ł́A^CX^vɊÂĒx⏞NCS̐݌vɂċc_B
@܂AxԂɉăQCؑւIuU[o(ؑւ^IuU[o)ĂBIuU[o̐݌v́Aؑւ𔺂덷n̈艻֋AB܂AWb^obt@p̂ɔׂĐ덷̎xPB
@ɁAlbg[Nł̒x⏞ɑ΂Ė@̗Lɂċc_Bł́A^CX^v̎ŗv鑗M̌vZ@̓ɂČAKvȐxœł邱ƂmFB܂ANCS̈ƂāAʐMH@ɂAƂƂ肠AĖ@ۂ̃lbg[Nœ삷邱Ƃɂ茟؂B

## 0661002 G Measuring and Characterizing Eye Movements for Performance Evaluation of Software Review i\tgEFAr[ɂ鐫\]̂߂̎vƓ́j

@\tgEFA̕iコ@ƂāAJɍ쐬镶ǂނƂŌso\tgEFAr[sĂB܂łɃ\tgEFAr[ɂs̌o⌟oir[\jコ邽߂ɁA܂܂ȃr[@₻̎xĂĂBAr[\ɂ́AJ҂̌l傫ȉe^ĂB̂߁Ar[@x̊Ĵ݂Ȃ炸Ar[ɂJ҂̍s𕪐͂Aǂ̂悤ȍsr[\ɉe^Ă邩𗝉邱ƂŁAJҋxsKvB{ł́AJ҂̍s̈Ⴂr[\isosojɂǂ̂悤ȉe^Ă邩𖾂炩ɂ邱ƂړIƂB r[sʓIɕ͂邽߂ɁA{ł̓r[ɂJ҂̎v@ĂBړsPʁAPʂŌv邱ƂŁA̓ǂݕƃr[\̊֌W𕪐͂邱Ƃ\ƂȂB{ł́Ar[ɂ鎋ړ(1)\es̊Ԃ̑JڂƁA(2)̕pr[ɂ镶Ԃ̑Jڂ2ɕނA烌r[\Ƃ̊֌W𕪐͂B
@܂Aes̊Ԃ̎Jڂ𕪐͂邽߂̎ƂāA\[XR[hpR[hr[ɂĎړƃr[\vB̌ʁAr[JnɃR[hŜTς悤ȓ"scan"̔팱҂ŌꂽBʓIȕ͂̌ʁAscan̎ԂZJ҂͕s̌oɎԂXꂽB̌ʂ́A]̃r[@ł͑ΏۂƂĂȂAsPʂł̓ǂݕ̈Ⴂr[\ɉe^Ă邱ƂĂB̂ƂAJ҂̋ɂāAr[JnɎԂăR[hŜTςƂA̓IȎw邱ƂŁAr[\߂邱Ƃł\B
@ɁAԂ̎Jڂ𕪐͂邽߂Ƀ\[XR[h̑ɁAvdl݌vpr[ɂ鎋ړƕs̌oEvsB̌ʁAr[Ώە̏ʕivdlڍא݌vj𒷎ԓǂłJ҂قǁAʓIɕsoĂXꂽBɁAR[hr[ɂĂ͗vdlƏڍא݌voX悭ǂłJ҂قǁA̕soĂXꂽB̌ʂ́AoIɌĂʕǂނƂ̗LpɂĒʓIȌʂ̂łAr[̋ɂLpȏ؋Ƃėp邱ƂłƍlB

## 0661007 Y@LI KIuU[opuVXeɊւ錤

@{bg}js[^puVXeł͒ʐMďʐMs߁AʐMxA萫A쐫̖肪B]̌͑쐫ɒڂƈ萫ɒڂƂɑʂA萫Ƒ쐫𗼗͏ȂB萫ɒڂƂāA󓮐̌⃍oXgp@B̕@͂ǂ̂悤ȒʐMxɑ΂Ăۏ؂邪AƊ̏𐳊mɑ҂ɓ邱ƂłBA쐫ɒڂƂāAƊfp@B̃VXe͍ƊVXeɑg݁A҂͍Ɗɉu}js[^𑀍삷邽ߒʐMx̉elȂƂǂB܂ARs[^OtBbNXpoI⏕ƊfɊÂ͒񎦂sƂŗǍDȑ쐫ێĂBȂÃVXe͗\ߍƊgݍނƂKvƂ邽߁AɌ덷݂ꍇɂ͏\Ȑ䐫\𔭊邱ƂłȂȂB܂A쐫ɒڂ͗_IɈ萫ۏ؂ĂȂꍇAs艻鋰ꂪBāAǍDȊ񎦂s߂ɂ͌ȈʒuԂƗ͏Ԃ̓AȂтɍƊ̍ƘcČ邱Ƃɂ͊o̒񎦂KvłBȂA]̌ł͂炪BĂȂB䂦A{ł͈ő쐫̗ǂVVXe\ѐ@ĂBĂ@́A]̍Ɗ̑g݌^uVXeɑ΂āAKIuU[oɂ鍄ƏڐG_̐ɂƊ̗͊o̍Čۏ؂B܂AƊ̓IȐsƂɂAf̌덷̉eBāAʒuƗ͂̃nCubh䑥̗̍pɂAȗ͏Ԃ̒Ǐ]ۏ؂邱Ƃő҂ɑ΂ČȊ񎦂sƂłBdvȖʂKIuU[o̐݌v@ƂāAQʂ̕@AɂLm߂B́A]炠郊Avmt֐p݌v@ŁAȗ͒Ǐ]1cmx̐xł̊񎦂\łAȒPȃQCݒŊ肨ѐsƂłB́AԎϊp@ŁAɂĔȃIuU[oɑ΂ĐȌ덷VXeołA䗝_x[XƂĉߓnAUȂǂRɐ݌v邱ƂłB܂Ax̔Ɗɑ΂ĂȔȂsƂłAŐmmx̐xŗǍDȍƊ񎦂łB̃{bg}js[^io|POjƗ́E[gZTpAzI1bԂ̒ʐMx𔭐AĂ̕@Ő݌vKIuU[opsAĎ@̗LmFB̌ʁAʐMxɂĂő쐫̗ǂuVXe\ł邱Ƃ킩B

## 0461002 rO SʃWt@C_ƑSʃJpf[^擾xɂ鉮O̎OfɊւ錤

@{_́AOz邽߂ɁASʃ[UWt@C_ƑSʃJp̗ǂf[^擾̎ƁȂSʃf[^̓ɂ鍂xȎOf̐Ɋւ錤łB
@LȓssOf́AssvAOn}Azš̗lXȃAvP[Vւ̉p҂ĂB̂߁A]J⃌[UWt@C_pĎ̂̎Of@̊JɍsĂBɃ[UWt@C_͕̂̌xɌvł邽ߍLȓss̃fɑpĂBA[UWt@C_vł̂̓[UƎ˂ꂽɌ肳邽߁ALȊŜ̃f𐶐ɂ́An_Ŏ擾Wf[^𓝍KvBɁÂ̐F̎擾ɂ̓JpKv邽߁A[UWt@C_pLȉO̎Ofɂ́A1)n_Ŏ擾Wf[^̍xȈʒuA2)v񐔂̍팸ړIƂIȌvʒǔA3)Wf[^ƃeNX̐mȓAۑƂȂB
@{\͂܂AO̑n_Ŏ擾Wf[^xɈʒu邽߂ɁA̕ǖʂ⓹HʂȂǂ̉Oɑ݂镽ʗ̈𗘗pʒu@ĂAV~[Vf[^ɂ鐸x]ƎŎ擾f[^ɓKpʂɂĎBɁAL悩Wt@C_i\ȗ̈悪鉮OɂāAIɃfΏ̖ۗ̈v팸邽߂̃f[^xVXeĂBVXéAv̍팸ƈʒû߂̕ʕ̎擾lČv̐xZoAfΏۗ̈̐x}bvWt@C_̑҂ɎƂŃf[^̎擾xBŌɁASʃWf[^ƑSʉ摜peNXtOf̐@ɂďqׁAŎ擾f[^琶OfBāA{̌ʂ𑍊AO̎Of̍̓W]ɂďqׂB

## 0661009 ㉪@ \ꂽwKyѐi邽߂̃G[WFg̓\̒

@{bgȂǂ̎VXemɑΉ邽߂ɂ́A@BwKiIvZ@Ƃ炩̍œK@ɂāAɉl邱ƂKvłB̂悤ȍœK@̓I֓Kpۂɂ͎@̑IłȂAړI֐A̍\A^p[^ȂǗlXȐ݌v肷KvB̏ꍇÂ悤Ȑ݌v͎҂̒ƎsɂčsB{\ł͓ (1)ړI֐̐݌vy鋭wK@ (2)iIvZɂ鐧̍\̎IȊl@ĂB
@ړI֐̐ݒɂāAɕ̖ړIlKvAӂɖړI֐߂邱ƂꍇB{ł́AɋwK𕡐̖ړIȂ鑽ړIœKɓKp邱ƂlBwK𑽖ړIœK֓Kpۂɂ́AړIƂ̕V̑킹ɂXJ[l̕VĂBA킹ɂ@ł͕VłƂ̍\܂pĂ炸AV̑傫ɕqȋwKVXeɂȂĂ܂Ƃ肪BŁA{ł͕V̍\ێAVl̑gݍ킹ɃoXgȋwK@Max-Min Actor-Critic (MMAC) ĂB MMAC͖ړIʂɗ^ꂽV֐Ƃɉl֐wK铯\̃W[AeԂɂŏ̉l֐ő剻łMax-MinœK𓾂BV~[VɂāAMax-MinœK͕V̘aő剻@ɔׂĊeVl̑gݍ킹̉e󂯂ɂƂBɂAړIœKɋwKKpۂɁAV֐̐݌veՂɂ邱ƂłƍlB
@̍\IɊl@ƂāAj[lbg[Nŕ\AiIvZ@ɂčœK NeuroEvolutionƌĂ΂@ɌĂB܂łɒĂĂNeuroEvolution@ 1̈qɑ΂1̌d݂蓖ĂĂ߁APʂ̒TsƂłȂB̂߁AK͂Gȍ\KvƂ֓KpꍇɁAǐߒɂĒTቺ肪BŁA{ł͉ߋɔ\1̃W[PʂƂėp\ɂ邱ƂŁAǐߒł͍\IȒT\ɂNeuroEvolution@ĂBĂI\ł̓j[Ԃ̌Ɏl̏d݂ƃlbg[N̕\̂ǂłʂȂ蓖Ă邱Ƃ\łBɂAlbg[NW[̑gݍ킹ŕ\łAPʂ̒TƃW[Pʂ̒TɈBV~[VɂAĎ@W[\𗘗p\TɂĔ}Rt邱ƂB

## 0661030 s@ Optimal Polymer Production in Continuous Polymerization Reactor

@A continuous stirred tank reactor (CSTR) is widely used to produce various polymers in the chemical industry. The same reactor is often operated at multiple operating conditions to produce several different grades of the same polymer according to the demand of customers. In this situation, not only the steady-state operation but also the grade transition operation plays an important role in the effective polymer production. In this thesis, the optimal polymer production in the CSTR is discussed from both the optimal steady-state operation and the optimal grade transition operation.
@First, a mathematical model is developed for describing the dynamic behavior of a general free radical polymerization in the CSTR. The weight-based molecular weight distribution (MWD) function and three parameters, the number-average molecular weight (Mn), weight-average molecular weight (Mw), and the polydispersity index (PDI) are proposed as the specification of polymer quality.
@Second, the optimal steady-state operating condition to produce polymers with the best match to a specified MWD profile is discussed. Through a study of typical types of free radical polymerization, we find that an operating point can be determined by specifying two MWD parameters, Mn and PDI when the termination by combination reaction is included. However, the simultaneous specification of Mn and PDI cannot determine an operating point when the termination reaction by combination is not included. In this case, we need to specify another parameter to determine an operating point. We also show that an appropriate objective function must be selected to determine an optimal operating condition by taking account of the relationship between the specified polymer quality parameters and decision variables.
@Finally, we discuss an optimal grade transition to minimize the raw material and energy costs during the grade transition operation as well as to shorten the transition time. We show that a combination of feed-forward and regulatory control system provide a good solution. AS a result, we can achieve optimal production of different grades of polymers by applying the steady state optimization and the optimal grade transition policy.

## 0661209 @ Making XML Database Systems Scalable to Computer Resources and Data Volumes

@Increasing use of XML has emphasized the need for scalable database systems that are capable of handling a large amount of XML data efficiently.
@This study explores effective methods for making a scalable XML database system in the following aspects: (a) scalability to data volumes, (b) scalable XML processing with a shared-nothing PC cluster, and (c) scalable database processing on shared-memory multiprocessors.
@In the study of (a), we propose an XQuery processing scheme in which an XML document is internally represented as a set of blocks and can directly be stored on secondary storage. Our experimental results showed that our storage scheme is scalable to data volumes and outperforms competing schemes with respect to I/O intensive workloads.
@In (b), we discuss on-the-fly XML processing using shared-nothing PC clusters. We propose a scheme for distributed and parallel query processing that employs a pass-by-reference semantics by using remote proxy. Previously proposed methods that use pass-by-value semantics have often suffered from redundant communication between processor elements and limited inter-operator parallelism. To cope with these problems, we developed a distributed XML query processing scheme that leverages the benefit of lazy evaluation. Our experimental results showed that our proposed scheme obtains up to 22x speedups compared with competitive methods, and demonstrated the importance of distributed XML database systems to take pass-by-reference semantics into consideration.
@In (c), we explain the internal locking in the buffer management module that prevents databases from being scalable to the number of processors. We further propose a scalable buffer management scheme that employs non-blocking synchronization instead of locking-based ones. Our experimental results revealed that our scheme can obtain nearly linear scalability to processors up to 64 processors, although the existing locking-based schemes do not scale beyond 16 processors.
@Finally, we conclude our studies with examining our XML native database system built on top of the three contributions.

## 0561032 {{ A Machine Learning Approach for Detecting Fraudulent Websites

@This dissertation presents phishing by using machine learning techniques. Phishing is a fraudulent activity defined as the acquisition of personal information by tricking an individual into believing the attacker is a trustworthy entity. Phishing attackers lure people by using phishing email'', as if it were sent by a legitimate corporation. The attackers also attract the email recipients into a phishing site'', which is the replica of an existing web page, to fool a user into submitting personal, financial, or password data.
@My motivation against phishing is supporting end users by informing that they are just visiting phishing sites. For doing so, I focus on improving the detection accuracy of the heuristics-based detection methods. Within the heuristics-based detection methods, several heuristics are used for calculating the likelihood of being a phishing site. The methods have a possibility to detect unreported phishing sites. The problems in the methods is that the detection accuracy is not so high. Accordingly, users would learn to distrust the system and would ignore the notification from detection systems.
@In the dissertation, I employ machine learning algorithms to improve the detection accuracy. At fist, I perform preliminary evaluation to emerge the issues on applying machine learning for combining heuristics. The result shows that AdaBoost, a typical one of machine learning, performs better than the traditional method. However, I observe that overfitting decreases the detection accuracy. In order to thwart the effectiveness of overfitting, I attempt to increase the number of URLs in my dataset by implementing a system which automatically collects and analyzes newly reported phishing sites.
@I then evaluate the performance for 9 machine learning algorithms, and the result shows that 8 out of 9 machine learning algorithms outperform the traditional method. Even if I modify the dataset which contains other phishing sites reported in different time period, or change the set of heuristics in detection, 7 out of 9 still perform better than the traditional method. Thus, I conclude that employing machine learning for detection of phishing sites is available.
@Based on the machine learning-based detection methods, I discuss how these methods adjust the detection strategies for each user. I also explain the techniques which aims to cover the weak point of human-being by using machine learning, named HumanBoost. The key idea of HumanBoost is to employ the user's trust decision as a new heuristic. My subject-within tests shows that the detection accuracy in the case of subjects with HumanBoost are higher in comparison among that in the cases of subjects without HumanBoost and the AdaBoost-based detection method. Finally, I propose HTTP Response Sanitizing which removes the input forms from the website. By comparing with the compulsory blocking, the loss of convenience would be lower in the case of my proposed method.
@In summary, this dissertation shows strategies of supporting end users to make trust decision, problems on detecting phishing sites, proposed machine learning-based detection methods for detection of phishing sites, and advanced technologies incorporate to machine learning-based detection methods. Finally, this dissertation has demonstrated that machine learning algorithms is feasible solution against phishing sites.

## 0361008 cij c Studies on Constructing, Refining and Exploiting Rich Information Resources

@ߔN̎Rꏈ́Aʂ̃R[pXɁAԑf͂W󂯉͂Ȃǂ̐󂢃x̏{ĊwKɗp悤ȓvI@SƂȂĂAӖt^ꂽR[pX̃bȏt^ꂽꎑɊÂ@͌ĂBA[ӖsȂ߂̊wKf[^ƂāA邢́A挒̂߂̒mƂāAbȏt^ꂽꎑ͏dvłB{\ł́Â悤ȁARꏈɂďdvȌꎑɏœ_āA̍\zABAp@ɂĔ\B
@ꎑƂẮANTTňȑOJĂIgW[({bn)Ap̎gƓp̑Ή֌WLqQp^[ΎA{ꎫ(Lexeed)AR[pX(w)ЉB{bńA{̖303000̈ӖNXɕނ̂łBQp^[Ύ́ApƖ̑gݍ킹LqAXɁAp̑Ή֌ŴłB Lexeed́ASɂāA{l̐l95͒mĂ{ƂđI肵A@BǂȗlXȏt^łBwR[pX́A\\AӖ\A̃R[pXɓIɕt^̂łB
@̌ꎑ̍\źAɐlłȂĂARXgƎԂBAjƂȂ镔쐬΁A͌IɊg邱ƂłBŁAꎑ̂AQp^[ΎɂČIg@ĂA̗L|^XNɂĕ]B
@XɁAӖȂǂ̃bȏƓv@Ƃ̗Z݁A̗L𒲍B{\ł́A\͌ʂ̐Iւ̗pāAlXȃẍӖ𗘗pA̗LB܂Ał͐lŕt^Ӗ𗘗p邪AӖ񎩑̂wKɂĊl@ɂāAB@ƂĒĂB ɂA\w񂾂łȂAӖpꏈ̗LAЂẮAꎑ\z@̏dvɂĂĊmFB

## 0661211 Yu Thomas Edison Chua Studies on Power, Thermal & False-path Aware Test Techniques for Modern System-on-Chips

@Rapid advances in semiconductor manufacturing processes and design tools have led to a relentless increase in chip complexity. High power consumption and heat densities have become major concerns. These problems are greatly exacerbated for System-on-Chips (SoCs) which integrate several functional cores on one chip. SoCs operating at multiple clock domains and very low power requirements are being utilized in the latest mobile devices. Thus, the testing of SoCs under power and temperature constraints have been rapidly gaining importance. For this thesis, we first introduce a novel method for designing power-aware test wrappers for embedded cores with multiple clock domains. By effectively partitioning the various clock domains, making use of bandwidth conversion, multiple shift frequencies and clock-gating, we gain greater flexibility in determining an optimal test schedule under very tight power constraints.
@For Socs, imposing power constraints does not always solve the problem of overheating due to the non-uniform distribution of power across the chip. We present two TAM/Wrapper co-design methodologies for SoCs that ensure thermal safety while still optimizing the test schedule. The methods combine simplified thermal-cost models with bin-partitioning and packing algorithms to minimize test time while satisfying temperature constraints.
@Another problem is the difficulty in identifying untestable multi-cycle paths. Their rapid and accurate identification could result in significant reductions in Automatic Test Pattern Generation (ATPG) time, tester memory, test cost and chip over-kill. For this, a novel method of identifying multi-cycle false paths at Register Transfer Level (RTL) is presented along with a case study to prove its effectiveness.

## 0661020 @ iq󑽓_U\ʋؓd}̎ԓIp^Pʓ@Ɋւ錤

@{_ł́Aiq󑽓_U\ʋؓd}ΏۂɁA^Pʊdʂ̋ԓIȓ̓c[ƂẴC^[tF[XJƁAVɒĂ3ev[gɂ銈dʂ̎ԓIp^Pʓ@ɂďqׂB
@^Pʂ͋؎kɂ^_oƋؐۂ琬A؎k𐧌䂷ŏ@\PʂłB؎k͉^_oɂ蔭dʂؐۏ邱ƂŋNB̎dʂ^PʊdʂƂԁBؒ͂́Adʂ̔ΕpxƁA^Pʂ̐ނω邱ƂŒ߂BdʂؐۂƂ̋ؐۓx͉^Pʂ̎ނɂقȂBΕpxx́A؎ؔJ̕]ɗpBA\ʋؓd}œM͔牺ŋԓIɕ̉^PʊdʂZꂽ̂ł̂ŁA̎wW𓾂邽߂ɂ͂܂X̉^Pʂ𓯒肷KvB
@ߔNA^Pʂ̓ɂ͑lŌvꂽ\ʋؓd}pBA]͕̉@ł͑lœԓIȏ\ɊpĂȂB{ł́A^Pʊdʂ̎ԓIȏɉ^Pʂ𓯒肷@ĂB܂Adɂ8~8̊iq64zuiq󑽓_\ʓdɂ쐬Bɋؔ͂56li8~7oɗUj\ʋؓd}𓯎ɌvłVXeJB
@ɁAꂽ\ʋؓd}̉̓c[ƂāAtopography}bvidʐ}j񎦂łC^[tF[X쐬B̃C^[tF[Xł́AdʂؐۂlqAj[Vŕ\邱ƂłB̂߁Adʂ̔΂⊱̏uԂ邢͒ŮdʂlqeՂɊώ@ł悤ɂȂB
@ŌɁAr񓪋؂̓ڐӎk20%MVC(Maximal Voluntary Contraction)A100%MVCABallistick3ނ̕\ʋؓd}pāA{@̓Kp\؂BāAꂼ̔ΕpxAxvł邱ƂBɁAdɗƋؐۑsɊpx^\ʋؓd}ɂKpAdɂ̔zƊpxؓd}ɑ΂Ă^Pʂ̓肪\ł邱ƂB
@{ŊJVXeɂ蓾^Pʂ̊ւ邢̎wẂAؔJ؋@\̒ʓI]ɗLpłƊ҂łB

## 0661015 ਈ q [U[̐̏p@BƂ̓IC^NVɊւ錤

@ߔNAЉ̏qE̐isɔAlԂ̐ԂŐlƋĕGȍƂsƂ̂ł@B⃍{bg̎v܂ĂB̂߂ɂ́A@B͊o^GoƂxȒmo\͂AlԂƓIEIȃC^NV\ł邱Ƃ߂B]AlԂƋ@BƂ̃C^NV͋@BɎtꂽeZT[̏ɊÂɂ̂SłBA̎@ł͔z⃁eiXAmCYAϋvȂǂ̖肩Aԉ𑜓xvRx߂邱ƂALȃC^NV̎ɂ͕s\łBŖ{ł́A@BƃC^NV郆[U[̐̏A^Cŋ@BɒʐM邱ƂɂAZT[Ȃ@BɉzIɃ[U[̗͊oGoAVm\{bg̐݌vAv[ĂB̓Iɂ́A@Bł̓[U[{bgɑ΂Ĕ̓xNg͓_̐sB{Av[͐ΏۂɈˑȂߕL{bgE@BɓKp邱ƂłB
@{Av[̗Lp؂邽߂ɁAZT[ȂYƗp@B}js[^A\ʋؓdiEMGjvuA[VLv[VXeȂVXe\zĎsB[U[̎ɊւEMGƎpA^Cŋ@BɒʐM邱Ƃŋ@BɉzIɗ͊o^GoAIIȃC^NVBXɁA[U[Ƌ@BAdʕ̎グEƂs^XNB
@āA{Av[ɔėpIɂ邽߂̌i߂B^XN̑Oɗ̓xNg肫\ߊwKĂ@ł́Aؓ̋p^[oIωɂA萸xĂ܂肪B̖邽߁A[U[̐̏ɉĕ֐P鋭wK̓ĂBwKp邱ƂɂāAiPj^XNsȂICŐ̃p[^̒\łAiQjZT̏o͓̖IȋtMȂĂwKiZTȂ{bgւ̓Kpj\łAƂ_BA̎ʂƋɁA{Av[̗_⍡̉pɂĂc_B

## 0661026 R{ K͂ȉzLԂl[UԃC^NV̕ył̎@

@ߔNẢzIȃIuWFNgzuzԂlXȗprɊp邽߂̌ɍsĂB̌́C̃[UŉzԂL鋤Lz(Virtual EnvironmentCVE) ƁC̃[UIuWFNg̓[VLvȂǂ̋ZpɂĂ荞݁AzԂɔf鉼z(Virtual RealityCVR) 2 ɕB
@ʂɁAVE ͑l̃[UΏۂƂ邽߁A\ȃT[oƂɌ傫ȒʐMш敝KvƂAVR ł́C̃[U̓荞ŉzԂɍ邽߂ɁA[VLvX[cSʃfBXvCAJACȂǁAȑugƂB̋Zpʃ[U̗lXȗprɉp邽߂ɂ́ARXgŎ邱ƂdvłB܂AVE,VR ZpZAlŋLԂɌẼIuWFNg̏荞߂VXe]܂ĂB{_ł́CVECVR ZpʓIȕyɂĎ邽߂̃t[[NAсA̎@Ɋւ鎟2 ̃gsbNɂĔ\B
@ɁARXĝT[orP2P ňʓIȃlbg[Nz(Networked Virtual EnvironmentCNVE) 邽߂ɁAlQ^IC[vCOQ[iȉAMMORPGjΏۂɁAP2P I[o[Clbg[NA[LeN̒ĂsBĎ@̐\mF邽߁ALAN œ삷vg^CvVXeɂƁAns-2 ɂV~[VCPlanetLab łWANł̃vg^CvVXeɂ̌ʁCĎ@̃lbg[NQ[ŁCem[ȟvZׁEʐMׂQ[VXẽXP[reBQ[vCɉeȂƁAʓIȃlbg[NɂāA镔̈̃vCxȂꍇłAʐMx\Ɏgp\Ȕ͈͂ɗ}邱Ƃł邱ƂmFB
@ɁAԂƉzԂ̊Ԃ̃C^NV邽߂̃t[[NƁA[UɁA[UɂƂĂdvȃIuWFNg荂pxōXV悤AR (eIuWFNg̈ʒuCȂǂ̍XV) ̔zpxQoS K@\ɂĂ̒ĂsBĂQoS K@\ɂĕ]邽߁AeIuWFNg̃f[^̍XV̕px̎ZB܂Avg^CvVXe쐬A[UړۂɎ̃IuWFNg̕\iK؂ɒ܂ł̎ԂlXȃIuWFNgɑ΂ČvB̌ʁAʏ̃C^[lbgі LAN ɂĒċ@\p\Ȑ\œ삷邱ƂmFB܂Az[Uɑ΂ẮAp\ȃf[^̍XVpxƒZxB邱ƂmFB

## 0661033 O Integrative analysis of transcriptomics and metabolomics in Escherichia coli

@In the era of post-genomics, a systematic and comprehensive understanding of the complex events of the organisms is a great concern in biology. Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR/MS) is the best MS technology for obtaining exact mass measurements owing to its great resolution and accuracy, and several outstanding FT-ICR/MS-based metabolomics approaches have been reported. In the present study, I proposed a procedure for metabolite annotation on direct-infusion FT-ICR/MS by taking into consideration the classification of metabolite-derived ions using correlation analyses. Integrated analysis based on information of isotope relations, fragmentation patterns by MS/MS analysis, co-occurring metabolites, and database searches (KNApSAcK and KEGG) can make it possible to annotate ions as metabolites and estimate cellular conditions based on metabolite composition. A total of 220 detected ions were classified into 174 metabolite derivative groups and 72 ions were assigned to candidate metabolites in the present work. Metabolic profiling has been able to distinguish between the growth stages with the aid of PCA. The constructed model using PLS regression for OD600 values as a function of metabolic profiles is very useful for identifying to what degree the ions contribute to the growth stages. Ten phospholipids which largely influence the constructed model are highly abundant in the cells. This approach can reveal that global modification of those phospholipids occurs as E. coli enters the stationary phase. Thus, the integrated approach involving correlation analyses, metabolic profiling, and database searching is efficient for high-throughput metabolomics. Furthermore, I performed the transition point analysis by applying the statistical method, Linear Dynamical System (LDS) to transcriptomics and metabolomics data, respectively and detected a time lag between transcriptional and metabolite levels. Finally, the integrative analysis of transcriptomics and metabolomics was performed based on gene-to-metabolite correlation analysis by taking into consideration a time lag.

## 0761203 @c A Proposal of a Practical Design of IP Mobility Implementation and Proof of its Validity through the live Internet as a Basement of the Future IP Mobility Activity

@Internet is growing by combining various communication media. In the future, it is considered the wireless connection, rapidly advancing recently, will be an important component as Internet connectivity. The research goal is to provide a practical implementation design of protocols that can support a huge number of mobile nodes and the mobile-centric operation style for the future. IPv6 and Mobile IPv6, the mobility protocol for IPv6, are the only realistic option to support that huge number of nodes. In this dissertation, the implementation design of the protocols is proposed, and verified by the actual implementation and operation. The validity and practicability of the stack has also been verified through several additional experiments of the IPv4 to IPv6 transition method, and the easy multihoming method both utilizing mobility features. Furthermore, deployment work to raise the availability of the infrastructure by publishing the stacks and starting service operation has also been done. Finally, by proposing and verifying a global mobility operation mechanism, this dissertation completes the proposal of a practical protocol implementation design for the future mobile-centric Internet infrastructure.

## 0661004 @ [UReLXgl񐄑EɊւ錤

@[U̚nDу[Ȕ ([UReLXgƂ) ɍ񋟂AReLXgˑ^񐄑EĂBReLXgˑ^񐄑E邽߂̋ZpIۑƂĈȉ̉ۑ肪B
@@(a) cȃACe̒烆[U̚nDу[UReLXgɍvACeI肷Zp̊mB
@@(b) cȐEACeɑ΂A[U̚nDуReLXgɊÂK؂ȃLO񋟂Zp̊mB
@@(c) ẼReLXgłȂAߋ/̃ReLXglACeI肷Zp̊mB
@{ł́A[UReLXgɈˑ[ŨACeɑ΂nDщlf\fĂ邱ƂɂAL̉ۑ (1)(3) Ɏg񂾁BȉA{ɂeۑɑ΂Av[ɂďqׂB
@]̏񐄑EZpł́A[ŨACeɑ΂D݂Ɋւf[^~ςÃf[^ɊÂă[UnDf쐬BA[U̚nD̓[UReLXgɉđlɕω邽߁AACeɑ΂D݂Ɋւf[^͂Ƃ邾ł͓K؂ȃ[UnD\邱Ƃ͂łȂBŁAۑ (1) ɑ΂ẮA[UReLXgɈˑ[UnD\AReLXgˑ^[UnDfĂBăf𗘗p邱ƂɂA[U̚nDу[UReLXgɍvACe𐄑EƂđI肷邱Ƃ\ƂB
@񐄑EZp̖ړÏƂāATsׂɂ郆[UŠyB[ȔTsׂɂ镉Sy邽߂ɁA[U֒񎦂鐄EACeȂ邱ƂdvłBŁAۑ (2) ɑ΂ẮAۑ (1) ɂđI肳ꂽEACeWɑ΂郉LO쐬Ȁn̃ACe[Uɒ񎦂@BɁA[UAĂReLXgɂāAACêǂ̑ɉlfsĂ̂\郆[ŨReLXgˑ^lffĂBăf𗘗p邱ƂɂAEACeWɑ΂A[UReLXgɈˑ[ỦlfɉLO[Uɒ񋟂邱Ƃ\ƂB
@[U̚nD́A[UĂ邻̎_ł̏󋵂łȂA[UߋɎs⏫\肵Ăsɂe󂯂ƍlB̂ɁAߋ/̃[U̎nIȍs[UReLXg̈ꕔłƍlBāAReLXgˑ^񐄑EɂāAߋ/̃[U̎nIȍslɓ邱Ƃ͏dvȉۑłBۑ (3) ɑ΂ẮA[U̍sf[^AẼReLXgɉ[U̍sp^[\AReLXgˑ^sfĂAăf𗘗p邱ƂɂAEReLXgɉAߋ/̎nIȃ[U̍sl񐄑E\ƂB
@{\ł́ALeۑɑ΂Av[ɂďqׁAeĕɂĐBɁǍʂAĕ̗LB

## 0661022 v Enhancing End-System Capabilities on the Internet with a Large-Scale Observational Approach iGhVXe̎xC^[lbg̑K͊ϑɊւ錤j

@{\ł́AC^[lbgɂVȃlbg[NṽfłAvP[Vw̃lbg[NvɂāÄӋƎɌg݂BAvP[Vw̃lbg[Nv́AKpeer-to-peerlbg[NAvP[V̂悤ɁAC^[lbg̃GhVXeœ삷AvP[V葽̃m[hƃlbg[Nɂ܂莩Iɓ삷悤ɂȂߒŕKvƂ悤ɂȂB̖ړÍAC^[lbg̍\vfim[h⃊NȂǁj̏Ԃ␫\E\lbg[NAAvP[Vɑ΂ĐvɂmɎW񋟂邱ƂłBAvP[V͒񋟂ꂽƂɁAg̃T[rẌێEgړIƂď̓肷邱ƂɂȂBȂ킿AAvP[Vw̃lbg[Nv́AGhVXe̎̔fޗ񋟂dvȎ葱łB
@{́Â悤ȃAvP[Vw̃lbg[NvC^[lbgŃT[rXƂēWJAv[ɂAlbg[NAvP[V_ɓłoƂړIƂAnIȒmTB̂߂̎g݂ƂāA{\ł͈ȉ̓eB܂A݂邠邢͌iK̃lbg[NAvP[V̎̎菇ӂ܂AAvP[Vw̃lbg[NvՁiapplication-oriented measurement platform, ȉ AOMPj̋@\v܂Ƃ߂B̏ŁAKw^peer-to-peerlbg[Nvlbg[NƂėpAOMP̎ɂāAlbg[Nł̉EK͊gȂǂl쌟،ʂƂɁA̎\ƎWJViIB܂AIPg|W̑K͒TɓK@̂ЂƂłDoubletreeۂAOMPɓڂߒAAOMP_ɑK͊ϑ@𓋍ڂł邱ƂmFBɍ\zIPg|WTVXeɂāȀ璷̒Ⴓϑm[ȟ̏ւ̑ϏQ̍ȂǂÂ悤ȑK͊ϑ@AvP[Vw̃lbg[Nv̗LȎi肤邱Ƃ咣BŌɁAۂ̃T[rXƂAOMPWJĂ߂́AЂĂ̓C^[lbgGhVXeɂƂĂ莩Rŏ_ȃvbgtH[ƂȂ邽߂̎wjƍ̉ۑɂĂ܂Ƃ߂B

## 0761209 c@ Shape Model Construction from Multi-View Range Images

@This thesis presents several new techniques related to geometric shape model construction from range images measured from multiple viewpoints. Most basic techniques required to solve this problem are registration and integration of multiple range images. We first present a method of robust registration of a pair of range images. For registering a pair of range images, the Iterative Closest Point (ICP) algorithm is widely used, but it is fragile to outliers due to occlusion and sensing errors. We integrated the least median of squares (LMedS) estimator with the ICP algorithm for making pairwise registration robust to outliers. This method is then extended for pairwise registration of a sequence of multiple range images.
@The procedures related to shape model generation from range images are summarized usually by a pipeline, and the input range images are first registered, then integrated as a geometric shape model. We propose to rearrange this pipeline. Registration of range images should be segmented into coarse and fine registration stages depending on existence of the initial registration.
@The coarse registration needed to be achieved without any knowledge of the initial registration. This problem is closely related to object recognition, and matching local invariant features is the most widely used approach. We propose a method for coarse registration of multiple range images by using the log-polar depth map for generating local invariant features used for correspondence establishment. In this method, point correspondence is certified by RANSAC algorithm, and registration tree is automatically generated by maximizing the number of inlier correspondences.
@The result of the coarse registration is used as the initial state for the fine registration. It is usually solved by minimization of the registration error starting from the initial state. We propose to solve fine registration and shape integration simultaneously. The input range images are first integrated, and each range image is registered to the integrated shape. Integration and registration are alternately iterated until the input shapes are well registered to the integrated shape. We use the signed distance field (SDF) for shape representation, and this representation is efficient for implementing this minimization process.
@In many cases of shape measurement of real objects, due to occlusion and object material properties, unmeasured regions remain as holes. We propose a method for filling holes of the integrated shape by iteratively fitting quadratic functions to the SDF. We also present that the differential geometry properties of the object surface can be extracted from the SDF. We present experimental results of the proposed methods applied on synthetic and real range images.

## 0561026 S The variational Bayesian approach to non-invasive brain imaging: MEG and fMRI

@{wʘ_ł͕ϕxCY@pMEGfMRIf[^͖̉@ĂBNP]C[WO̓qg̔]@\ѐ_o\𒲂ׂ邽߂ɌƂ̂łȂZpłBƂɔ]}iMEGjƋ@\IjC摜ifMRIj́A悭pC[WO@łA]ɂĖLxȃf[^񋟂ĂB
@͂߂MEGዅA[t@NgϕxCY@BMEG͊ዅ^R̗]ȎiዅA[t@NgjɕqłAꂪዅ^̐Ȃ^XNMEG̓Kp͈͂肵ĂBĕ@͊ዅA[t@NǧƂčE̊ዅɂꂼP_C|[肵AdƔ]d𓯎ɂ邱ƂŃA[`t@NgsB
@Ɏoh񎦂ƂfMRIf[^摜̊𒊏oϕxCY։͖@ĂB̕@fMRIf[^ƂɑΉh摜ABԂ̃W[\ʂĂȂ킹B̉BꃂW[\p邱ƂɂāAfMRIf[^݂̂珉߂Č摜č\ł悤ɂȂB
@ŌɁA̓̎@̊֘AɂďqׂB̒Ď@͔NP]C[WOɂ_o@̉𖾂ɂ炽ȓ؂JƍlB
@ Ȋw U