Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
AWS で加速する機械学習 / Accelerate Machine Learning wit...
Search
Hokuto Hoshi
March 12, 2018
Technology
1.1k
1
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
AWS で加速する機械学習 / Accelerate Machine Learning with AWS
Hokuto Hoshi
March 12, 2018
More Decks by Hokuto Hoshi
See All by Hokuto Hoshi
しぶいSRE: サーバから見えない障害にどう向き合うか。ラストワンマイルのデバッグ実践 / Shibui SRE
kanny
15
8.5k
AIに "賭ける": LayerX の AI エージェント基盤と活用最前線 / Building an AI Agent Platform at LayerX
kanny
3
560
AIとともに歩む情報セキュリティ / Information Security with AI
kanny
5
4.9k
開発も運用もビジネス部門も! クラウドで実現する「つらくない」統制とセキュリティ / Effortless Governance and Security Enabled by the Cloud
kanny
5
5k
転生CISOサバイバル・ガイド / CISO Career Transition Survival Guide
kanny
4
2.7k
Connecting organisation with Technology
kanny
0
390
Why Slack - 5 years of Cookpad with Slack
kanny
0
210
Security by builders - セキュリティ監視をクラウドで「つくる」 / Security by builders
kanny
7
2.8k
セキュリティ担当者から見た re:Invent と AWS Security Hub / Impression of re:Invent and AWS Security Hub
kanny
2
4.3k
Other Decks in Technology
See All in Technology
Master Dataグループ紹介資料
sansan33
PRO
1
4.8k
名刺メーカーDevグループ 紹介資料
sansan33
PRO
0
1.2k
Goでデータパイプラインを作ろう
sansantech
PRO
1
420
【CEDEC2026】ゲームシナリオライターを支援するAIツール開発の実践 ― 設計とプロンプトの工夫 ―
cygames
PRO
1
790
研究開発部の紹介 / Sansan R&D Profile
sansan33
PRO
4
24k
MIRU 2026 チュートリアル
keisuke198619
0
890
20分でわかるセキュアAPI
nwiizo
2
250
ブラウザ研修 2026
recruitengineers
PRO
6
930
TypeScript入門 2026
recruitengineers
PRO
3
550
20260804_Q4AzureUpdateBite_FabricDataAgentの精度を高める設計.pdf
matayuuu
1
130
猫付きpingコマンドを自作
uyuki234
0
210
Contract One Engineering Unit 紹介資料
sansan33
PRO
0
19k
Featured
See All Featured
Noah Learner - AI + Me: how we built a GSC Bulk Export data pipeline
techseoconnect
PRO
0
350
Bridging the Design Gap: How Collaborative Modelling removes blockers to flow between stakeholders and teams @FastFlow conf
baasie
0
640
Code Review Best Practice
trishagee
74
20k
Balancing Empowerment & Direction
lara
6
1.2k
The Power of CSS Pseudo Elements
geoffreycrofte
82
6.5k
DevOps and Value Stream Thinking: Enabling flow, efficiency and business value
helenjbeal
1
320
Future Trends and Review - Lecture 12 - Web Technologies (1019888BNR)
signer
PRO
0
3.7k
How to train your dragon (web standard)
notwaldorf
97
6.7k
VelocityConf: Rendering Performance Case Studies
addyosmani
332
25k
Sam Torres - BigQuery for SEOs
techseoconnect
PRO
0
460
Kristin Tynski - Automating Marketing Tasks With AI
techseoconnect
PRO
0
440
The Organizational Zoo: Understanding Human Behavior Agility Through Metaphoric Constructive Conversations (based on the works of Arthur Shelley, Ph.D)
kimpetersen
PRO
0
400
Transcript
"84ͰՃ͢Δػցֶश ΫοΫύουגࣜձࣾ ేછ୩༔Ұ "844UBSUVQ%BZ5PLZP
ే ΄͠΄͘ͱ !LBOJ@C ΫοΫύουגࣜձࣾ ΠϯϑϥετϥΫνϟʔ෦෦ 43& ηΩϡϦςΟΤϯδχΞ
"84ೝఆ4" %FW0QTΤϯδχΞ 1SPGFTTJPOBM
None
Ϩγϐ ສҎ্ ࠃͷ݄ؒར༻ऀ ສਓ
ରԠݴޠ ݴޠΧࠃ ւ֎ͷ݄ؒར༻ऀ ສਓҎ্
ΫοΫύουͱ"84 ʹ%$͔ΒશҠߦ BQOPSUIFBTU͓ΑͼVTFBTU Λओʹར༻ άϩʔόϧαʔϏεओʹ64ͰՔಇ "NB[PO&$Πϯελϯε d
ఔ "NB[PO4ΦϒδΣΫτ ԯҎ্ SFRTҎ্ ϐʔΫ࣌ಉ࣌ϦΫΤετ
IUUQTTQFBLFSEFDLDPNLBOOZNJBPKJBOTIVNPGBMTFSPHVXPJJHBO[JOJTVSVBLJUFLVUJZB IUUQTTQFBLFSEFDLDPNBZFNPTNBDIJOFMFBSOJOHPQTBUDPPLQBE
"84SF*OWFOU IUUQTXXXTMJEFTIBSFOFU"NB[PO8FC4FSWJDFTDPODPOUBJOFSJ[FENBDIJOFMFBSOJOHPOBXT
"84ͰՃ͢Δػցֶश
ྉཧ͖Ζ͘ ΈࠐΈχϡʔϥϧωοτϫʔΫ ʹΑΔྉཧը૾ͷࣗಈೝࣝ εϚʔτϑΥϯͷࣸਅͷɺ ྉཧࣸਅΛࣗಈతʹऩूه
%&.0
None
৽͍͠ઓ ॳͷػցֶशϓϩμΫτ ηϛϦΞϧλΠϜͳը૾ೝࣝ 8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ ඞཁͩͬͨͷ
৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ ࣦഊͯ͠طଘͷڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠ڥ ૉૣ͘ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧڥ
৽͍͠ઓ ॳͷػցֶशϓϩμΫτ ηϛϦΞϧλΠϜͳը૾ೝࣝ 8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ ඞཁͩͬͨͷ
৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ ࣦഊͯ͠طଘͷڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠ڥ ૉૣ͘ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧڥ ৽͍͠αʔϏε࡞ΓΛ lՃz͢Δڥ
εέʔϥϏϦςΟͷ֬อ ༧ଌͰ͖ͳ͍τϥϑΟοΫ େྔͷը૾ΞοϓϩʔυͳͲ ಥવ57ʹհ͞ΕΔ͜ͱ͋Δ͔͠Εͳ͍ ࣮ࡍ͋ͬͨ
ը૾ͷΞοϓϩʔυఆʹ͕͔͔࣌ؒΔ dඦNT ಉظॲཧͯ͠͠·͏ͱϢʔβମݧΛѱ͘͢Δ
ඇಉظͳΞʔΩςΫνϟ ը૾Ξοϓϩʔυͱը૾ఆΛݸผʹॲཧ ͦΕͧΕΛεέʔϧ͍͢͠Α͏ͳߏʹ "NB[PO4ͷQSFTJHOFE63-ͰΞοϓϩʔυ 4OPUJpDBUJPOΛར༻͠ॲཧ͖͢ը૾ͷใ "NB[PO424
ڥͷ ͜Ε·Ͱͷ8FCαʔόڥͱҧ͏ڥ͕ٻΊΒΕͨ ҧ͏ݴޠ ҧ͏ϫʔΫϩʔυ ҧ͏ϋʔυΣΞ (16
ίϯςφڥΛར༻ ϥϯλΠϜ(16υϥΠόɺઃఆΛ "NB[PO&$4Λར༻ͯ͠ཧ طʹଟ͘ͷαʔϏε͕ίϯςφڥͰՔಇ
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ {
bucket: thumbnails key: 123.jpg }
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ {
bucket: thumbnails key: 123.jpg } ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ {
bucket: thumbnails key: 123.jpg } ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμϯϩʔυ
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ {
bucket: thumbnails key: 123.jpg } ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμϯϩʔυ ը૾ͷఆΛߦ͍݁ՌΛ "1*αʔόʹ௨
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ { bucket: thumbnails key: 123.jpg }
ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμϯϩʔυ ը૾ͷఆΛߦ͍݁ՌΛ "1*αʔόʹ௨ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ ಛผͳίʔυΠϯϑϥΛ༻ҙͤͣ εέʔϥϒϧͳߏʹ
API Server "1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ {
bucket: thumbnails key: 123.jpg } αϜωΠϧͷμϯϩʔυ ը૾ͷఆΛߦ͍݁ՌΛ "1*αʔόʹ௨ ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster ECS Cluster 424ͷΩϡʔʹԠͯ͡ &$4λεΫεέʔϧ
"1*αʔόʹ 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕ 424ʹೖ͞ΕΔ { bucket: thumbnails
key: 123.jpg } ϝοηʔδΛ 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμϯϩʔυ ը૾ͷఆΛߦ͍݁ՌΛ "1*αʔόʹ௨ API Server API Server &$4λεΫ͔Βͷఆ݁ՌΛ͚ͭͩ
&$4্Ͱͷࣸਅఆ ࣸਅఆʹ(16Λ༻͍ͯ͠Δ ఆॲཧʹ(16ΞΫηϥϨʔγϣϯ͕ར༻Ͱ͖Δ ಉՁ֨ଳͷ$16Πϯελϯεͱൺͯdഒͷੑೳࠩ HΠϯελϯεͰࣸਅఆ༻ͷ&$4$MVTUFSΛߏ
IBLPʹΑΔΞϓϦέʔγϣϯఆٛ IBLPIUUQTHJUIVCDPNFBHMFUNUIBLP &$4ޓͷίϯςφσϓϩΠπʔϧ ZBNMKTPOOFUܗࣜͰͷΞϓϦέʔγϣϯఆٛ "VUP4DBMFɺ-#ͷࢦఆͳͲՄೳ
શ։ൃऀ͕ࣗͰఆٛΛॻ͍ͯσϓϩΠՄೳ scheduler: type: ecs region: ap-northeast-1 cluster: hako-production-g2 desired_count: 1 app: image: food-photo-classifier cpu: 128 memory: 3072 memory_reservation: 2048 env: AWS_REGION: ap-northeast-1 COOKPADNET_ENV: production ...
ӡ༻࣮ ը૾ೝࣝϞσϧΛҰΞοϓσʔτ ΞοϓσʔτΓӽ͑ɺେ͖ͳͳ͘ӡ༻Ͱ͖͍ͯΔ ສਓҎ্ ͷྉཧࣸਅΛه ͷϢʔβʔ͕ར༻ ສຕҎ্
݄ݱࡏ
৽͍͠ઓ ॳͷػցֶशϓϩμΫτ ηϛϦΞϧλΠϜͳը૾ೝࣝ 8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ ඞཁͩͬͨͷ
৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ✅ طଘͷڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠ڥ✅ ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧڥ
࣮ݧڥͮ͘Γ खΛಈ͔ͯ͠ػցֶशͷݟΛಘΔඞཁ͕͋ͬͨ "84ڥશͯΠϯϑϥ෦͕ཧ͍͕ͯͨ͠ʜ Ϧιʔεͷ༻ҙ͕͍͔ͭͳ͍ ͍ΘΏΔຊ൪ڥͷӡ༻ʹ߹Θͳ͍
ٻΊΒΕΔͷαʔϏεʹඞཁͳ࣮ݧ݁Ռ தԝཧʹݶք͕๚Εͨ ֤ࣄۀ෦ Πϯϑϥ෦ ٕज़ج൫ܥ෦ॺ ݚڀ ։ൃ෦
ݚڀ։ൃج൫ ʮ͔ͤͬ͘৮ΕΔͷ͔ͩΒউखʹਐΊΒΕΕྑ͍ʯ ݚڀ։ൃ෦ͷΤϯδχΞ͕ࣗ༝ʹखΛಈ͔ͤΔॴ ຊ൪ڥͱ͞Ε͍ͯΔ ཧ͖͢෦͚͓ͩ͑͞ɺଟ͘Λ։ൃऀҠৡ
ΞΧϯτ "84ΞΧϯτΛผʹ࡞ͯ͠ "840SHBOJ[BUJPOΛར༻ͯ͠ଞͷ༻్ʹ࡞ *".6TFSຊ൪ڥͷݸਓϢʔβ͔Β"TTVNF3PMFPS440 $MPVE5SBJM$POpHΛ༗ޮʹͯ͠PVUQVUઌΛอޢ
lى͜Δલʹ͙zͰͳ͘l͙͢ؾ͚ͮΔzlޙ͔Β͑Δz
ωοτϫʔΫڥ ౿Έ44)αʔόຊ൪ΞΧϯτͷ71$ Πϯϑϥ෦ཧ 71$1FFSJOHͰଓͯ͠ू /BNFλάΛͬͨ%/4ਖ਼Ҿ͖ɺٯҾ͖Λఏڙ "84$POpH3VMFTͰηΩϡϦςΟάϧʔϓͷΠϯλʔωοτ
ղ์ΛνΣοΫ
None
ίετཧ ࡉ͔͍ཧ͠ͳ͍͕͋ΔఔѲ ৵ೖͳͲͷΞϥʔτʹͳΔ #VEHFUTػೳΛར༻ ༧ଌֹͳͲͰ4/4/PUJpDBUJPOΛड͚औΓରԠ
None
αϙʔτͷ׆༻ σϕϩούʔαϙʔτ͔ΒΤϯλʔϓϥΠζαϙʔτͷҠߦ ࣌ؒͱࣝΛങ͏ ։ൃऀ͕ࣗͰΛղܾͰ͖ΔΑ͏ʹ ར༻نʹΑͬͯϏδωεͰͳ͍ͷͰཁݟੵΓ
։ൃऀΞΧϯτ ݚڀ։ൃ༻ͱಉ༷͕ͩશ։ൃऀ͕ࣗ༝ʹར༻Մೳ ৭ʑͳ"84αʔϏεΛձࣾ࣋ͪͰݕূͰ͖Δ
ར༻࣮ "84ΛࣗͰ৮ͬͯΈΔਓ͕ͱͯ૿͑ͨ ༷ʑͳ"84αʔϏεͷར༻ͱऔΓࠐΈ "1*(BUFXBZ -BNCEB $PHOJUP %ZOBNP%#
&MFNFOUBM.FEJB 4FSWJDF (MVF ,JOFTJT 1PMMZ 4BHF.BLFS 93BZ -FY 1JOQPJOU "QQ4ZODFUDʜ ৽ػೳɺαʔϏεͷग़ΛՃ ྉཧ͖Ζ͘ "MFYB4LJMM DPPLQBE57 FUD
ֶΜͩ͜ͱ ৽͍͜͠ͱΛૉૣ͘খ࢝͘͞ΊΔͨΊʹݖݶͱ͕ཁΔ ҆৺ͯࣗ͠༝ʹ͑Δڥͮ͘Γ͕ͦ͜ॏཁ "84ʹ࣮ݱ͢ΔͨΊͷπʔϧ͕ἧ͍ͬͯΔ
Ͳ͏͍ͬͯͨͷ͔ʁ
ࣗݾհ છ୩༔Ұ<:VJDIJSP4PNFZB> ΫοΫύουגࣜձࣾݚڀ։ൃ෦ ϦαʔνΤϯδχΞ݄d ʰػցֶशʹඞཁͳج൫Λ "84্ʹߏங͢Δʱ UXJUUFSDPN!BZFNPT@Z
HJUIVCDPNBZFNPT IUUQTXXXBZFNPTNF
εϞʔϧελʔτͰεέʔϧ͢Δ (16࣮ݧڥΛ"84্ʹߏங͢Δ
എܠΫοΫύουݚڀ։ൃ෦ ݄ʹൃ ॳظϝϯόʔ໊ ݄ݱࡏ ࠃʹ໊ ւ֎ʹ໊
ෳͷਂֶशϓϩδΣΫτ (16࣮ݧڥ͕ඞཁ ϢʔβʔϑΟʔυόοΫ ͷࣗಈλά͚ ը૾ੳʹΑΔϨγϐͷྨ Ϩγϐࡐྉ໊ͷࣗಈਖ਼نԽ
(16࣮ݧڥͷཁ݅ ෳਓ͕ෳͷ࣮ݧΛߦ͏͜ͱ͕ग़དྷΔ
&$ΠϯελϯεΛ࣮ݧڥͱͯ͠͏ ΦϯσϚϯυΠϯελϯεͷར͕͋Δ σϕϩούʔλεΫʹԠͯ͡ඞཁͳͱεϖοΫ͕มԽ͢Δ 1/7*%*"(16࠷৽Ϟσϧͷαϙʔτ ઐ༗Πϯελϯεͷར
(16λεΫجຊతʹഉଞͰɺଞͷ։ൃऀͱಉ࣌ʹར༻ͮ͠Β͍ ݚڀ։ൃతͰɺࡶʹTVEPग़དྷΔಠཱͨ͠ڥ͕ศར
(16࣮ݧڥ &$ ͕ग़དྷΔ·Ͱ &$ΠϯελϯεΛ্ཱͪ͛ɺTTIΛૄ௨ͤ͞Δ ඞཁͳιϑτΣΞͷΠϯετʔϧʗઃఆ OWJEJBESJWFS $6%"
DV%//ͳͲ݁ߏେม ".* "NB[PO.BDIJOF*NBHF ͷར༻
ιϑτΣΞɺϛυϧΣΞͷ Πϯετʔϧͱઃఆ ".*࡞ͱཧ
ιϑτΣΞɺϛυϧΣΞͷ Πϯετʔϧͱઃఆ εφοϓγϣοτ ".*ͷ࡞ ".*࡞ͱཧ
ιϑτΣΞɺϛυϧΣΞͷ Πϯετʔϧͱઃఆ ίʔυԽͯ͠όʔδϣϯཧ εφοϓγϣοτ ".*ͷ࡞ ".*࡞ͱཧ
IUUQTEPDTBXTBNB[PODPNKB@KQEMBNJMBUFTUEFWHVJEFXIBUJTEMBNJIUNM "84%FFQ-FBSOJOH".* ਂֶशʹ ຆͲͷ߹ڞ௨ͯ͠ ඞཁͳιϑτΣΞ͕ηοτ Ξοϓ͞ΕͨϚωʔδυͳ".* l/7*%*"$6%"ɺ/7*%*"DV%//ɺ͓Αͼ࠷ਓؾ͕ߴ͍σΟʔϓϥʔχϯά ϑϨʔϜϫʔΫͷ࠷৽ϦϦʔεΛ༻͢ΔΑ͏ʹࣄલʹઃఆ͞Εͯఏڙ͞Ε·͢ɻz
ࣗ࡞".* ίʔυԽͯ͠όʔδϣϯཧ εφοϓγϣοτ ".*ͷ࡞ "84 %FFQ-FBSOJOH ".* FD3VO*OTUBODFT
ࣗ࡞".* ίʔυԽͯ͠όʔδϣϯཧ εφοϓγϣοτ ".*ͷ࡞ "84 %FFQ-FBSOJOH ".* FD3VO*OTUBODFT ".*Λར༻ͯ͠ɺڥΛؚΊͯΦϯσϚϯυͳ
(16࣮ݧڥΛ࣮ݱ
".*Λར༻͢Δ͜ͱʹΑΓιϑτΣΞͷΠϯετʔϧͱઃఆ ࡞ۀΛল͘͜ͱ͕ग़དྷͨ 71$ 4FDVSJUZ(SPVQͳͲͷઃఆ߲͕͍ͬͯΔ $IBUCPUӽ͠ʹطఆΛར༻ͯ͠ΠϯελϯεΛ࡞Δ ͏ͻͱ ʙ$IBUCPU
"84-BNCEBʹΑΔ&$ͷૢ࡞ʙ
͏ͻͱ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ
DSFBUFXPSLCFODI 0VUHPJOH8FCIPPLT ͏ͻͱ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ
DSFBUFXPSLCFODI 0VUHPJOH8FCIPPLT TMBDL@VJE6 DPNNBOEDSFBUFXPSLCFODI MBNCEB*OWPLF FD3VO*OTUBODFT 71$ 4VCOFU /BNFλά
ڥಛ༗ͷઃఆ͜͜ͰΔ ͏ͻͱ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ
*ODPNJOH8FCIPPLT ATTIBZFNPTXPSLCFODIEOTDPNQBOZA TTI ͏ͻͱ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ
͞Βʹ͏ͻͱ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷࢹʙ ͍ͬͯͳ͍Πϯελϯεͷམͱ͠ΕͳͲ͕ൃੜ අ༻໘Ͱେ͖ͳͰͳ͍͕ɺ͍ۙকདྷ৫֦େͷোนʹ ͳΓ͔Ͷͳ͍ ར༻୯Ձͷߴ͍Πϯελϯεʹ͍ͭͯɺΞΠυϧঢ়ଶΛࢹ
͞Βʹ͏ͻͱ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷࢹʙ ;[[
͞Βʹ͏ͻͱ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷࢹʙ IPVSMZ FD4UPQ*OTUBODFT ;[[
͞Βʹ͏ͻͱ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷࢹʙ IPVSMZ FD4UPQ*OTUBODFT ;[[ ར༻ऀʹ௨ʗࣗಈͰ4UPQ
(16࣮ݧڥ·ͱΊ ".*ͷར༻ʹΑΔଈ࣌ར༻Մೳͳ &$ΠϯελϯεΛ4MBDL͔Βૢ࡞͢Δ ΠϯελϯεΛ্ཱͪ͛ΔࡍͷઃఆΛڞ௨Խ 4MBDLͷ*ODPNJOH8FCIPPL0VUHPJOH8FCIPPLͱ "NB[PO"1*(BUFXBZ"84-BNCEBͷΈ߹Θͤ
҆ՁͰαʔόʔϨεͳ$IBU#PU͕࣮Ͱ͖Δ
࣮ݧ͔Βຊ൪
ػցֶशϞσϧͷ։ൃͱσϓϩΠ ػցֶशϞσϧσϓϩΠ͞ΕͯॳΊͯՁΛੜΉ FHྉཧ͖Ζ͘ ϞσϧͱΞϓϦέʔγϣϯຖʹઃܭͱσϓϩΠ༻։ൃ࡞ۀ͕ੜ͡Δ ʮػցֶशΞϓϦέʔγϣϯઃܭ։ൃ୭͕Δͷ͔ʁʯ
εϞʔϧελʔτ৫ͷεέʔϥϏϦςΟʹڞʹࢧো͕ग़Δ
"NB[PO4BHF.BLFS "84ͷϚωʔδυػցֶशϓϥοτϑΥʔϜ "84SF*OWFOUͰൃද ػցֶशϞσϧͷֶशͱσϓϩΠΛαϙʔτ ϓϥοτϑΥʔϜͤͰεέʔϥϒϧͳσϓϩΠख๏
ϞχλϦϯά"#ςετͳͲϓϥοτϑΥʔϜͳΒͰͷ ػೳࠓޙʹظ
·ͱΊ
͘͢खΛಈ͔͠αʔϏεΛͭ͘ΔͨΊʹ ηϧϑαʔϏεԽ શͯΛཧ͢ΔͷͰͳ͘ τϨʔεͰ͖ΔΈͮ͘Γ $MPVE5SBJM $POGJH ΤϯλʔϓϥΠζαϙʔτͷ׆༻ ྉཧ͖Ζ͘ͷެ։
(16࣮ݧڥ ".*Λར༻ͨ͠ (16࣮ݧڥͷඋ ".* Πϯελϯεͷ མͱ͠Εࢹ Πϯελϯεཧ$IBU#PU -BNCEB $MPVE8BUDI ΦϯσϚϯυͳ(16࣮ݧڥΛ࣮ݱ
8F`SFIJSJOH IUUQTDPPLQBEKPCT